2021
Pan, Yanghe; Ni, Jianbing; Su, Zhou
FL-PATE: Differentially Private Federated Learning with Knowledge Transfer Inproceedings
In: 2021 IEEE Global Communications Conference (GLOBECOM), pp. 1-6, 2021.
Abstract | Links | BibTeX | Tags:
@inproceedings{9685079,
title = {FL-PATE: Differentially Private Federated Learning with Knowledge Transfer},
author = {Yanghe Pan and Jianbing Ni and Zhou Su},
url = {https://ieeexplore.ieee.org/abstract/document/9685079},
doi = {10.1109/GLOBECOM46510.2021.9685079},
year = {2021},
date = {2021-12-11},
urldate = {2021-01-01},
booktitle = {2021 IEEE Global Communications Conference (GLOBECOM)},
pages = {1-6},
abstract = {Federated learning provides a solution for data privacy protection, while enabling training over the local data samples, without exchanging them. However, it is far from practical and secure because data privacy is still vulnerable due to the well-studied attacks, e.g., membership inference attacks and model inversion attacks. In this paper, to further prevent data leakage against these attacks, we propose FL-PATE, a differentially private federated learning framework with knowledge transfer. Specifically, participants with sensitive data are grouped to train teacher models under federated learning settings, and the knowledge of teacher models is transferred to a publicly accessible student model for prediction via aggregating teacher models' outputs of public datasets. A modified client-level differential privacy mechanism is used to guarantee each participant's data privacy during the corresponding teacher model's training process. The proposed framework preserves participant's privacy against membership inference attacks and the differential privacy cost is fixed. The privacy analysis and experiments demonstrate that trained teacher and student models have an excellent performance in accuracy and robustness theoretically and empirically.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Federated learning provides a solution for data privacy protection, while enabling training over the local data samples, without exchanging them. However, it is far from practical and secure because data privacy is still vulnerable due to the well-studied attacks, e.g., membership inference attacks and model inversion attacks. In this paper, to further prevent data leakage against these attacks, we propose FL-PATE, a differentially private federated learning framework with knowledge transfer. Specifically, participants with sensitive data are grouped to train teacher models under federated learning settings, and the knowledge of teacher models is transferred to a publicly accessible student model for prediction via aggregating teacher models' outputs of public datasets. A modified client-level differential privacy mechanism is used to guarantee each participant's data privacy during the corresponding teacher model's training process. The proposed framework preserves participant's privacy against membership inference attacks and the differential privacy cost is fixed. The privacy analysis and experiments demonstrate that trained teacher and student models have an excellent performance in accuracy and robustness theoretically and empirically.
Patricia N. Pena Michel R. C. Alves, Karen Rudie
Discrete-event systems subject to unknown sensor attacks Bachelor Thesis
2021.
Abstract | Links | BibTeX | Tags:
@bachelorthesis{Alves2021,
title = {Discrete-event systems subject to unknown sensor attacks},
author = {Michel R. C. Alves, Patricia N. Pena, Karen Rudie},
url = {https://link.springer.com/article/10.1007/s10626-021-00351-3#citeas},
doi = {10.1007/s10626-021-00351-3},
year = {2021},
date = {2021-12-10},
urldate = {2021-12-10},
journal = {Discrete Event Dynamic Systems},
abstract = {This work is set in the context of supervisory control of discrete-event systems under partial observation. Attackers that are able to insert or erase occurrences of particular output symbols can tamper with the supervisor’s observation and by doing so, can lead the controlled system to undesirable states. We consider a scenario with multiple attackers, each one being an element of a set, called the attack set. We also assume that only one of the attackers within an attack set is acting, although we don’t know which one. According to previous results in the literature, a supervisor that enforces a given legal language, regardless of which attacker is acting, can be designed if the legal language is controllable and satisfies a property called P-observability for an attack set. The latter is an extended notion of observability and is related with the supervisor’s ability to always distinguish between outputs that require different control actions, even if the outputs were attacked. We present a new approach for checking if a given language is P-observable for an attack set, by first introducing a visual representation as well as some definitions that capture the attack’s effect. Additionally, we present two algorithms that together allow us to verify if a given language is P-observable for an attack set, when it is represented as an automaton.},
keywords = {},
pubstate = {published},
tppubtype = {bachelorthesis}
}
This work is set in the context of supervisory control of discrete-event systems under partial observation. Attackers that are able to insert or erase occurrences of particular output symbols can tamper with the supervisor’s observation and by doing so, can lead the controlled system to undesirable states. We consider a scenario with multiple attackers, each one being an element of a set, called the attack set. We also assume that only one of the attackers within an attack set is acting, although we don’t know which one. According to previous results in the literature, a supervisor that enforces a given legal language, regardless of which attacker is acting, can be designed if the legal language is controllable and satisfies a property called P-observability for an attack set. The latter is an extended notion of observability and is related with the supervisor’s ability to always distinguish between outputs that require different control actions, even if the outputs were attacked. We present a new approach for checking if a given language is P-observable for an attack set, by first introducing a visual representation as well as some definitions that capture the attack’s effect. Additionally, we present two algorithms that together allow us to verify if a given language is P-observable for an attack set, when it is represented as an automaton.
Kolahdouzi, Mojtaba; Sepas-Moghaddam, Alireza; Etemad, Ali
Face Trees for Expression Recognition Miscellaneous
2021.
@misc{kolahdouzi2021face,
title = {Face Trees for Expression Recognition},
author = {Mojtaba Kolahdouzi and Alireza Sepas-Moghaddam and Ali Etemad},
url = {https://arxiv.org/pdf/2112.02487.pdf},
year = {2021},
date = {2021-12-05},
urldate = {2021-01-01},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Wang, Yuntao; Su, Zhou; Ni, Jianbing; Zhang, Ning; Shen, Xuemin
Blockchain-Empowered Space-Air-Ground Integrated Networks: Opportunities, Challenges, and Solutions Journal Article
In: IEEE Communications Surveys Tutorials, pp. 1-1, 2021.
@article{9631953,
title = {Blockchain-Empowered Space-Air-Ground Integrated Networks: Opportunities, Challenges, and Solutions},
author = {Yuntao Wang and Zhou Su and Jianbing Ni and Ning Zhang and Xuemin Shen},
doi = {10.1109/COMST.2021.3131711},
year = {2021},
date = {2021-12-01},
urldate = {2021-01-01},
journal = {IEEE Communications Surveys Tutorials},
pages = {1-1},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Edraki, Amin; Chan, Wai-Yip; Jensen, Jesper; Fogerty, Daniel
Speech Intelligibility Prediction Using Spectro-Temporal Modulation Analysis Journal Article
In: IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 29, pp. 210-225, 2021.
@article{9269417,
title = {Speech Intelligibility Prediction Using Spectro-Temporal Modulation Analysis},
author = {Amin Edraki and Wai-Yip Chan and Jesper Jensen and Daniel Fogerty},
doi = {10.1109/TASLP.2020.3039929},
year = {2021},
date = {2021-11-24},
journal = {IEEE/ACM Transactions on Audio, Speech, and Language Processing},
volume = {29},
pages = {210-225},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Siahkouhi, Mohammad; Razaqpur, Ghani; Hoult, N. A.; Baghban, Mohammad Hajmohammadian; Jing, Guoqing
Utilization of carbon nanotubes (CNTs) in concrete for structural health monitoring (SHM) purposes: A review Journal Article
In: Construction and Building Materials, vol. 309, pp. 125137, 2021, ISSN: 0950-0618.
Abstract | Links | BibTeX | Tags: Carbon Nanotube, Cement composite, Concrete, Infrastructure, Self-healing, Self-sensing property, Structural health monitoring
@article{SIAHKOUHI2021125137,
title = {Utilization of carbon nanotubes (CNTs) in concrete for structural health monitoring (SHM) purposes: A review},
author = {Mohammad Siahkouhi and Ghani Razaqpur and N. A. Hoult and Mohammad Hajmohammadian Baghban and Guoqing Jing},
url = {https://www.sciencedirect.com/science/article/pii/S0950061821028816},
doi = {https://doi.org/10.1016/j.conbuildmat.2021.125137},
issn = {0950-0618},
year = {2021},
date = {2021-11-21},
urldate = {2021-01-01},
journal = {Construction and Building Materials},
volume = {309},
pages = {125137},
abstract = {Carbon nanotubes (CNT) as a functional filler can increase the electrical conductivity property of concrete and thus provide intrinsic self-sensing properties with no need for external sensors to monitor the behavior of concrete infrastructure and structures containing CNT. CNT-cement composites also improve mechanical strength and have higher energy absorption capacity. The sensitivity of their electrical conductivity to external physical parameters, such as strain, stress, load, temperature, displacement, and pressure, makes them suitable for structural health monitoring (SHM) applications. This paper presents a comprehensive review of the CNT properties, fabrication process, composition, and sensing characteristics as well as challenges for applying CNT concrete as part of a self-sensing structure. Furthermore, the self-healing property of CNT, as an integral feature of future smart concrete infrastructure is discussed.},
keywords = {Carbon Nanotube, Cement composite, Concrete, Infrastructure, Self-healing, Self-sensing property, Structural health monitoring},
pubstate = {published},
tppubtype = {article}
}
Carbon nanotubes (CNT) as a functional filler can increase the electrical conductivity property of concrete and thus provide intrinsic self-sensing properties with no need for external sensors to monitor the behavior of concrete infrastructure and structures containing CNT. CNT-cement composites also improve mechanical strength and have higher energy absorption capacity. The sensitivity of their electrical conductivity to external physical parameters, such as strain, stress, load, temperature, displacement, and pressure, makes them suitable for structural health monitoring (SHM) applications. This paper presents a comprehensive review of the CNT properties, fabrication process, composition, and sensing characteristics as well as challenges for applying CNT concrete as part of a self-sensing structure. Furthermore, the self-healing property of CNT, as an integral feature of future smart concrete infrastructure is discussed.
Sarkar, Pritam; Etemad, Ali
Self-Supervised Audio-Visual Representation Learning with Relaxed Cross-Modal Temporal Synchronicity Miscellaneous
2021.
Abstract | Links | BibTeX | Tags:
@misc{sarkar2021selfsupervised,
title = {Self-Supervised Audio-Visual Representation Learning with Relaxed Cross-Modal Temporal Synchronicity},
author = {Pritam Sarkar and Ali Etemad},
url = {https://arxiv.org/pdf/2111.05329.pdf},
year = {2021},
date = {2021-11-14},
urldate = {2021-01-01},
abstract = {We present CrissCross, a self-supervised framework for learning audio-visual representations. A novel notion is introduced in our framework whereby in addition to learning the intra-modal and standard 'synchronous' cross-modal relations, CrissCross also learns 'asynchronous' cross-modal relationships. We show that by relaxing the temporal synchronicity between the audio and visual modalities, the network learns strong time-invariant representations. Our experiments show that strong augmentations for both audio and visual modalities with relaxation of cross-modal temporal synchronicity optimize performance. To pretrain our proposed framework, we use 3 different datasets with varying sizes, Kinetics-Sound, Kinetics-400, and AudioSet. The learned representations are evaluated on a number of downstream tasks namely action recognition, sound classification, and retrieval. CrissCross shows state-of-the-art performances on action recognition (UCF101 and HMDB51) and sound classification (ESC50). The codes and pretrained models will be made publicly available. },
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
We present CrissCross, a self-supervised framework for learning audio-visual representations. A novel notion is introduced in our framework whereby in addition to learning the intra-modal and standard 'synchronous' cross-modal relations, CrissCross also learns 'asynchronous' cross-modal relationships. We show that by relaxing the temporal synchronicity between the audio and visual modalities, the network learns strong time-invariant representations. Our experiments show that strong augmentations for both audio and visual modalities with relaxation of cross-modal temporal synchronicity optimize performance. To pretrain our proposed framework, we use 3 different datasets with varying sizes, Kinetics-Sound, Kinetics-400, and AudioSet. The learned representations are evaluated on a number of downstream tasks namely action recognition, sound classification, and retrieval. CrissCross shows state-of-the-art performances on action recognition (UCF101 and HMDB51) and sound classification (ESC50). The codes and pretrained models will be made publicly available.
Phinnemore, Rachel; Cimolino, Gabriele; Sarkar, Pritam; Etemad, Ali; Graham, T. C. Nicholas
Happy Driver: Investigating the Effect of Mood on Preferred Style of Driving in Self-Driving Cars Inproceedings
In: Proceedings of the 9th International Conference on Human-Agent Interaction, pp. 139–147, Association for Computing Machinery, Virtual Event, Japan, 2021, ISBN: 9781450386203.
Abstract | Links | BibTeX | Tags: driver mood, driving styles, Self-driving cars, user experience.
@inproceedings{10.1145/3472307.3484169,
title = {Happy Driver: Investigating the Effect of Mood on Preferred Style of Driving in Self-Driving Cars},
author = {Rachel Phinnemore and Gabriele Cimolino and Pritam Sarkar and Ali Etemad and T. C. Nicholas Graham},
url = {https://doi.org/10.1145/3472307.3484169},
doi = {10.1145/3472307.3484169},
isbn = {9781450386203},
year = {2021},
date = {2021-11-09},
urldate = {2021-01-01},
booktitle = {Proceedings of the 9th International Conference on Human-Agent Interaction},
pages = {139–147},
publisher = {Association for Computing Machinery},
address = {Virtual Event, Japan},
series = {HAI '21},
abstract = {Self-driving cars are around the corner, yet little is known about how users of self-driving cars will react to the car’s driving style, and whether the driver’s mood affects their driving style preference. This paper explores the impact of users’ mood on driving style preference in self-driving cars. An experiment was conducted online (N=182) to investigate participants’ preference for three driving styles (conservative, moderate, aggressive) under three induced moods (calm, neutral, excited). Measures of arousal, valence, and driving satisfaction were recorded. Overall, participants scored the aggressive driving style lowest, irrespective of driver mood. Participants’ mood impacted preference, where a mismatch between driving style and mood induced prior to the driving style predicted lower driver satisfaction scores. We conclude with the design recommendation that driving styles in self-driving cars should not be overly aggressive, and drivers’ mood should be taken into consideration when designing driving styles.},
keywords = {driver mood, driving styles, Self-driving cars, user experience.},
pubstate = {published},
tppubtype = {inproceedings}
}
Self-driving cars are around the corner, yet little is known about how users of self-driving cars will react to the car’s driving style, and whether the driver’s mood affects their driving style preference. This paper explores the impact of users’ mood on driving style preference in self-driving cars. An experiment was conducted online (N=182) to investigate participants’ preference for three driving styles (conservative, moderate, aggressive) under three induced moods (calm, neutral, excited). Measures of arousal, valence, and driving satisfaction were recorded. Overall, participants scored the aggressive driving style lowest, irrespective of driver mood. Participants’ mood impacted preference, where a mismatch between driving style and mood induced prior to the driving style predicted lower driver satisfaction scores. We conclude with the design recommendation that driving styles in self-driving cars should not be overly aggressive, and drivers’ mood should be taken into consideration when designing driving styles.
Li, Dayu; Zhu, Xiaodan; Li, Yang; Wang, Suge; Li, Deyu; Liao, Jian; Zheng, Jianxing
Emotion Inference in Multi-Turn Conversations with Addressee-Aware Module and Ensemble Strategy Inproceedings
In: Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pp. 3935–3941, Ässociation for Computational Linguistics, Önline and Punta Cana, Dominican Republic, 2021.
Abstract | Links | BibTeX | Tags:
@inproceedings{li-etal-2021-emotion,
title = {Emotion Inference in Multi-Turn Conversations with Addressee-Aware Module and Ensemble Strategy},
author = {Dayu Li and Xiaodan Zhu and Yang Li and Suge Wang and Deyu Li and Jian Liao and Jianxing Zheng},
url = {https://aclanthology.org/2021.emnlp-main.320},
year = {2021},
date = {2021-11-01},
urldate = {2021-11-01},
booktitle = {Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing},
pages = {3935--3941},
publisher = {Ässociation for Computational Linguistics},
address = {Önline and Punta Cana, Dominican Republic},
abstract = {Emotion inference in multi-turn conversations aims to predict the participant's emotion in the next upcoming turn without knowing the participant's response yet, and is a necessary step for applications such as dialogue planning. However, it is a severe challenge to perceive and reason about the future feelings of participants, due to the lack of utterance information from the future. Moreover, it is crucial for emotion inference to capture the characteristics of emotional propagation in conversations, such as persistence and contagiousness. In this study, we focus on investigating the task of emotion inference in multi-turn conversations by modeling the propagation of emotional states among participants in the conversation history, and propose an addressee-aware module to automatically learn whether the participant keeps the historical emotional state or is affected by others in the next upcoming turn. In addition, we propose an ensemble strategy to further enhance the model performance. Empirical studies on three different benchmark conversation datasets demonstrate the effectiveness of the proposed model over several strong baselines.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Emotion inference in multi-turn conversations aims to predict the participant's emotion in the next upcoming turn without knowing the participant's response yet, and is a necessary step for applications such as dialogue planning. However, it is a severe challenge to perceive and reason about the future feelings of participants, due to the lack of utterance information from the future. Moreover, it is crucial for emotion inference to capture the characteristics of emotional propagation in conversations, such as persistence and contagiousness. In this study, we focus on investigating the task of emotion inference in multi-turn conversations by modeling the propagation of emotional states among participants in the conversation history, and propose an addressee-aware module to automatically learn whether the participant keeps the historical emotional state or is affected by others in the next upcoming turn. In addition, we propose an ensemble strategy to further enhance the model performance. Empirical studies on three different benchmark conversation datasets demonstrate the effectiveness of the proposed model over several strong baselines.
Shi, Zhan; Liu, Hui; Min, Martin Renqiang; Malon, Christopher; Li, Li Erran; Zhu, Xiaodan
Retrieval, Analogy, and Composition: A framework for Compositional Generalization in Image Captioning Inproceedings
In: Findings of the Association for Computational Linguistics: EMNLP 2021, pp. 1990–2000, Ässociation for Computational Linguistics, Punta Cana, Dominican Republic, 2021.
Abstract | Links | BibTeX | Tags:
@inproceedings{shi-etal-2021-retrieval-analogy,
title = {Retrieval, Analogy, and Composition: A framework for Compositional Generalization in Image Captioning},
author = {Zhan Shi and Hui Liu and Martin Renqiang Min and Christopher Malon and Li Erran Li and Xiaodan Zhu},
url = {https://aclanthology.org/2021.findings-emnlp.171},
year = {2021},
date = {2021-11-01},
urldate = {2021-11-01},
booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2021},
pages = {1990--2000},
publisher = {Ässociation for Computational Linguistics},
address = {Punta Cana, Dominican Republic},
abstract = {Image captioning systems are expected to have the ability to combine individual concepts when describing scenes with concept combinations that are not observed during training. In spite of significant progress in image captioning with the help of the autoregressive generation framework, current approaches fail to generalize well to novel concept combinations. We propose a new framework that revolves around probing several similar image caption training instances (retrieval), performing analogical reasoning over relevant entities in retrieved prototypes (analogy), and enhancing the generation process with reasoning outcomes (composition). Our method augments the generation model by referring to the neighboring instances in the training set to produce novel concept combinations in generated captions. We perform experiments on the widely used image captioning benchmarks. The proposed models achieve substantial improvement over the compared baselines on both composition-related evaluation metrics and conventional image captioning metrics.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Image captioning systems are expected to have the ability to combine individual concepts when describing scenes with concept combinations that are not observed during training. In spite of significant progress in image captioning with the help of the autoregressive generation framework, current approaches fail to generalize well to novel concept combinations. We propose a new framework that revolves around probing several similar image caption training instances (retrieval), performing analogical reasoning over relevant entities in retrieved prototypes (analogy), and enhancing the generation process with reasoning outcomes (composition). Our method augments the generation model by referring to the neighboring instances in the training set to produce novel concept combinations in generated captions. We perform experiments on the widely used image captioning benchmarks. The proposed models achieve substantial improvement over the compared baselines on both composition-related evaluation metrics and conventional image captioning metrics.
He, Weinan; Huang, Canming; Liu, Yongmei; Zhu, Xiaodan
WinoLogic: A Zero-Shot Logic-based Diagnostic Dataset for Winograd Schema Challenge Inproceedings
In: Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pp. 3779–3789, Ässociation for Computational Linguistics, Önline and Punta Cana, Dominican Republic, 2021.
Abstract | Links | BibTeX | Tags:
@inproceedings{he-etal-2021-winologic,
title = {WinoLogic: A Zero-Shot Logic-based Diagnostic Dataset for Winograd Schema Challenge},
author = {Weinan He and Canming Huang and Yongmei Liu and Xiaodan Zhu},
url = {https://aclanthology.org/2021.emnlp-main.307},
year = {2021},
date = {2021-11-01},
urldate = {2021-11-01},
booktitle = {Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing},
pages = {3779--3789},
publisher = {Ässociation for Computational Linguistics},
address = {Önline and Punta Cana, Dominican Republic},
abstract = {The recent success of neural language models (NLMs) on the Winograd Schema Challenge has called for further investigation of the commonsense reasoning ability of these models. Previous diagnostic datasets rely on crowd-sourcing which fails to provide coherent commonsense crucial for solving WSC problems. To better evaluate NLMs, we propose a logic-based framework that focuses on high-quality commonsense knowledge. Specifically, we identify and collect formal knowledge formulas verified by theorem provers and translate such formulas into natural language sentences. Based on these true knowledge sentences, adversarial false ones are generated. We propose a new dataset named WinoLogic with these sentences. Given a problem in WinoLogic, NLMs need to decide whether the plausible knowledge sentences could correctly solve the corresponding WSC problems in a zero-shot setting. We also ask human annotators to validate WinoLogic to ensure it is human-agreeable. Experiments show that NLMs still struggle to comprehend commonsense knowledge as humans do, indicating that their reasoning ability could have been overestimated.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
The recent success of neural language models (NLMs) on the Winograd Schema Challenge has called for further investigation of the commonsense reasoning ability of these models. Previous diagnostic datasets rely on crowd-sourcing which fails to provide coherent commonsense crucial for solving WSC problems. To better evaluate NLMs, we propose a logic-based framework that focuses on high-quality commonsense knowledge. Specifically, we identify and collect formal knowledge formulas verified by theorem provers and translate such formulas into natural language sentences. Based on these true knowledge sentences, adversarial false ones are generated. We propose a new dataset named WinoLogic with these sentences. Given a problem in WinoLogic, NLMs need to decide whether the plausible knowledge sentences could correctly solve the corresponding WSC problems in a zero-shot setting. We also ask human annotators to validate WinoLogic to ensure it is human-agreeable. Experiments show that NLMs still struggle to comprehend commonsense knowledge as humans do, indicating that their reasoning ability could have been overestimated.
Salari, O.; Zaad, K. Hashtrudi; Bakhshai, A.; Jain, P.
Pseudo Multi Level Space Vector Modulation Technique for Multi Source Inverters Inproceedings
In: 2021 IEEE Energy Conversion Congress and Exposition (ECCE), pp. 4979-4986, 2021.
@inproceedings{9595582,
title = {Pseudo Multi Level Space Vector Modulation Technique for Multi Source Inverters},
author = {O. Salari and K. Hashtrudi Zaad and A. Bakhshai and P. Jain},
doi = {10.1109/ECCE47101.2021.9595582},
year = {2021},
date = {2021-11-01},
urldate = {2021-11-01},
booktitle = {2021 IEEE Energy Conversion Congress and Exposition (ECCE)},
pages = {4979-4986},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Chawla, Ishan; Pathak, P M; Notash, Leila; Samantaray, A K; Li, Qingguo; Sharma, U K
Workspace analysis and design of large-scale cable-driven printing robot considering cable mass and mobile platform orientation Journal Article
In: Mechanism and Machine Theory, vol. 165, pp. 104426, 2021, ISSN: 0094-114X.
Abstract | Links | BibTeX | Tags: 3D printing, Cable-driven parallel robots, Design analysis, Workspace analysis
@article{CHAWLA2021104426,
title = {Workspace analysis and design of large-scale cable-driven printing robot considering cable mass and mobile platform orientation},
author = {Ishan Chawla and P M Pathak and Leila Notash and A K Samantaray and Qingguo Li and U K Sharma},
url = {https://www.sciencedirect.com/science/article/pii/S0094114X21001841},
doi = {https://doi.org/10.1016/j.mechmachtheory.2021.104426},
issn = {0094-114X},
year = {2021},
date = {2021-11-01},
journal = {Mechanism and Machine Theory},
volume = {165},
pages = {104426},
abstract = {This paper proposes a methodology to determine the printable workspace of a cable robot by taking cable mass and mobile platform orientation into account. This workspace has a characteristic feature of avoiding the collision of lower cables with the previous layers of the structure during 3D printing while maintaining cable tensions inside the prescribed positive tension bounds. Since redundant robot is taken into consideration, the printable workspace possesses multiple solutions and hence, a unique solution needs to be selected based upon a selection criterion. Therefore, to select a unique solution, this paper proposes a selection criterion to minimize cable sagging and compares it with the conventional selection criterion of minimizing cable tension. For comparative analysis, various factors such as printable workspace, cable tensions and platform deflection of the robot are considered. The proposed selection criterion is further utilized to study the impact of variation in the geometric parameters on the printable workspace and the capacity margin of the robot. Finally, an optimal design is proposed using genetic algorithm for the desired printable workspace as well as constraint minimum capacity margin with an objective to minimize the installation space of the cable robot.},
keywords = {3D printing, Cable-driven parallel robots, Design analysis, Workspace analysis},
pubstate = {published},
tppubtype = {article}
}
This paper proposes a methodology to determine the printable workspace of a cable robot by taking cable mass and mobile platform orientation into account. This workspace has a characteristic feature of avoiding the collision of lower cables with the previous layers of the structure during 3D printing while maintaining cable tensions inside the prescribed positive tension bounds. Since redundant robot is taken into consideration, the printable workspace possesses multiple solutions and hence, a unique solution needs to be selected based upon a selection criterion. Therefore, to select a unique solution, this paper proposes a selection criterion to minimize cable sagging and compares it with the conventional selection criterion of minimizing cable tension. For comparative analysis, various factors such as printable workspace, cable tensions and platform deflection of the robot are considered. The proposed selection criterion is further utilized to study the impact of variation in the geometric parameters on the printable workspace and the capacity margin of the robot. Finally, an optimal design is proposed using genetic algorithm for the desired printable workspace as well as constraint minimum capacity margin with an objective to minimize the installation space of the cable robot.
Shen, Shuaiqi; Yu, Chong; Zhang, Kuan; Ni, Jianbing; Ci, Song
Adaptive and Dynamic Security in AI-Empowered 6G: From an Energy Efficiency Perspective Journal Article
In: IEEE Communications Standards Magazine, vol. 5, no. 3, pp. 80-88, 2021.
@article{9579394,
title = {Adaptive and Dynamic Security in AI-Empowered 6G: From an Energy Efficiency Perspective},
author = {Shuaiqi Shen and Chong Yu and Kuan Zhang and Jianbing Ni and Song Ci},
doi = {10.1109/MCOMSTD.101.2000090},
year = {2021},
date = {2021-10-18},
urldate = {2021-01-01},
journal = {IEEE Communications Standards Magazine},
volume = {5},
number = {3},
pages = {80-88},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Cimolino, Gabriele; Askari, Sussan; Graham, T. C. Nicholas
Beyond Fun: Players’ Experiences of Accessible Rehabilitation Gaming for Spinal Cord Injury Inproceedings
In: The 23rd International ACM SIGACCESS Conference on Computers and Accessibility, Association for Computing Machinery, Virtual Event, USA, 2021, ISBN: 9781450383066.
Abstract | Links | BibTeX | Tags: Accessibility, Digital games, Rehabilitation, Spinal cord injury
@inproceedings{10.1145/3441852.3471227,
title = {Beyond Fun: Players’ Experiences of Accessible Rehabilitation Gaming for Spinal Cord Injury},
author = {Gabriele Cimolino and Sussan Askari and T. C. Nicholas Graham},
url = {https://doi.org/10.1145/3441852.3471227},
doi = {10.1145/3441852.3471227},
isbn = {9781450383066},
year = {2021},
date = {2021-10-17},
urldate = {2021-10-17},
booktitle = {The 23rd International ACM SIGACCESS Conference on Computers and Accessibility},
publisher = {Association for Computing Machinery},
address = {Virtual Event, USA},
series = {ASSETS '21},
abstract = {Rehabilitation gaming—play of digital games that incorporate rehabilitation exercises—is
a well-known and broadly applicable way to make physical rehabilitation more fun.
It can motivate patients with spinal cord injury to engage in exercises that they
find boring and can be as effective as traditional physiotherapy. However, patients’
needs are not only physical. Rehabilitation also needs to help patients overcome the
psychological trauma of spinal cord injury. For patients coping with disability, hopelessness,
depression, anxiety, or a loss of identity, rehabilitation gaming may provide benefits
beyond making exercise more fun. We asked six participants with spinal cord injury
to play three cycling-based rehabilitation games to determine how play might change
their experiences of rehabilitation. They said that rehabilitation games may be able
to help patients to actively participate in their rehabilitation, help them to rediscover
who they are, and show them a better future living with spinal cord injury.},
keywords = {Accessibility, Digital games, Rehabilitation, Spinal cord injury},
pubstate = {published},
tppubtype = {inproceedings}
}
Rehabilitation gaming—play of digital games that incorporate rehabilitation exercises—is
a well-known and broadly applicable way to make physical rehabilitation more fun.
It can motivate patients with spinal cord injury to engage in exercises that they
find boring and can be as effective as traditional physiotherapy. However, patients’
needs are not only physical. Rehabilitation also needs to help patients overcome the
psychological trauma of spinal cord injury. For patients coping with disability, hopelessness,
depression, anxiety, or a loss of identity, rehabilitation gaming may provide benefits
beyond making exercise more fun. We asked six participants with spinal cord injury
to play three cycling-based rehabilitation games to determine how play might change
their experiences of rehabilitation. They said that rehabilitation games may be able
to help patients to actively participate in their rehabilitation, help them to rediscover
who they are, and show them a better future living with spinal cord injury.
a well-known and broadly applicable way to make physical rehabilitation more fun.
It can motivate patients with spinal cord injury to engage in exercises that they
find boring and can be as effective as traditional physiotherapy. However, patients’
needs are not only physical. Rehabilitation also needs to help patients overcome the
psychological trauma of spinal cord injury. For patients coping with disability, hopelessness,
depression, anxiety, or a loss of identity, rehabilitation gaming may provide benefits
beyond making exercise more fun. We asked six participants with spinal cord injury
to play three cycling-based rehabilitation games to determine how play might change
their experiences of rehabilitation. They said that rehabilitation games may be able
to help patients to actively participate in their rehabilitation, help them to rediscover
who they are, and show them a better future living with spinal cord injury.
Moulton, Richard Hugh; Hamgini, Behnam Behinaein; Khouzani, Zahra Abedi; Meira-Gòes, Rômulo; Wang, Fei; Rudie, Karen
Using Subobservers to Synthesize Opacity-Enforcing Supervisors Miscellaneous
2021.
Abstract | Links | BibTeX | Tags:
@misc{moulton2021using,
title = {Using Subobservers to Synthesize Opacity-Enforcing Supervisors},
author = {Richard Hugh Moulton and Behnam Behinaein Hamgini and Zahra Abedi Khouzani and Rômulo Meira-Gòes and Fei Wang and Karen Rudie},
url = {https://arxiv.org/pdf/2110.04334.pdf},
year = {2021},
date = {2021-10-08},
urldate = {2021-10-08},
abstract = {In discrete-event system control, the worst-case time complexity for computing a system's observer is exponential in the number of that system's states. This results in practical difficulties since some problems require calculating multiple observers for a changing system, e.g., synthesizing an opacity-enforcing supervisor. Although calculating these observers in an iterative manner allows us to synthesize an opacity-enforcing supervisor and although methods have been proposed to reduce the computational demands, room exists for a practical and intuitive solution. Here we extend the subautomaton relationship to the notion of a subobserver and demonstrate its use in reducing the computations required for iterated observer calculations. We then demonstrate the subobserver relationship's power by simplifying state-of-the-art synthesis approaches for opacity-enforcing supervisors under realistic assumptions. },
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
In discrete-event system control, the worst-case time complexity for computing a system's observer is exponential in the number of that system's states. This results in practical difficulties since some problems require calculating multiple observers for a changing system, e.g., synthesizing an opacity-enforcing supervisor. Although calculating these observers in an iterative manner allows us to synthesize an opacity-enforcing supervisor and although methods have been proposed to reduce the computational demands, room exists for a practical and intuitive solution. Here we extend the subautomaton relationship to the notion of a subobserver and demonstrate its use in reducing the computations required for iterated observer calculations. We then demonstrate the subobserver relationship's power by simplifying state-of-the-art synthesis approaches for opacity-enforcing supervisors under realistic assumptions.
Muise, Christian; Belle, Vaishak; Felli, Paolo; McIlraith, Sheila; Miller, Tim; Pearce, Adrian R.; Sonenberg, Liz
Efficient Multi-agent Epistemic Planning: Teaching Planners About Nested Belief Miscellaneous
2021.
@misc{muise2021efficient,
title = {Efficient Multi-agent Epistemic Planning: Teaching Planners About Nested Belief},
author = {Christian Muise and Vaishak Belle and Paolo Felli and Sheila McIlraith and Tim Miller and Adrian R. Pearce and Liz Sonenberg},
url = {https://arxiv.org/pdf/2110.02480.pdf
},
year = {2021},
date = {2021-10-06},
urldate = {2021-10-06},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Cimolino, Gabriele; Askari, Sussan; Graham, T. C. Nicholas
The Role of Partial Automation in Increasing the Accessibility of Digital Games Journal Article
In: Proc. ACM Hum.-Comput. Interact., vol. 5, no. CHI PLAY, 2021.
Abstract | Links | BibTeX | Tags: Artificial Intelligence, Automation, game accessibility, personalization, shared control
@article{10.1145/3474693,
title = {The Role of Partial Automation in Increasing the Accessibility of Digital Games},
author = {Gabriele Cimolino and Sussan Askari and T. C. Nicholas Graham},
url = {https://doi.org/10.1145/3474693},
doi = {10.1145/3474693},
year = {2021},
date = {2021-10-01},
urldate = {2021-10-01},
journal = {Proc. ACM Hum.-Comput. Interact.},
volume = {5},
number = {CHI PLAY},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
abstract = {Digital games are designed to be controlled using hardware devices such as gamepads,
keyboards, and cameras. Some device inputs may be inaccessible to players with motor
impairments, rendering them unable to play. Games and devices can be adapted to enable
play, but for some players these adaptations may not go far enough. Games may require
inputs that some players cannot provide with any device. To address this problem,
we introduce partial automation, an accessibility technique that delegates control
of inaccessible game inputs to an AI partner. Partial automation complements and builds
on other approaches to improving games' accessibility, including universal design,
player balancing, and interface adaptation. We have demonstrated partial automation
in two games for the rehabilitation of spinal cord injury. Six study participants
with vastly different motor abilities were able to play both games. Participants liked
the increased personalization that partial automation affords, although some participants
were confused by aspects of the AI's behaviour.},
keywords = {Artificial Intelligence, Automation, game accessibility, personalization, shared control},
pubstate = {published},
tppubtype = {article}
}
Digital games are designed to be controlled using hardware devices such as gamepads,
keyboards, and cameras. Some device inputs may be inaccessible to players with motor
impairments, rendering them unable to play. Games and devices can be adapted to enable
play, but for some players these adaptations may not go far enough. Games may require
inputs that some players cannot provide with any device. To address this problem,
we introduce partial automation, an accessibility technique that delegates control
of inaccessible game inputs to an AI partner. Partial automation complements and builds
on other approaches to improving games' accessibility, including universal design,
player balancing, and interface adaptation. We have demonstrated partial automation
in two games for the rehabilitation of spinal cord injury. Six study participants
with vastly different motor abilities were able to play both games. Participants liked
the increased personalization that partial automation affords, although some participants
were confused by aspects of the AI's behaviour.
keyboards, and cameras. Some device inputs may be inaccessible to players with motor
impairments, rendering them unable to play. Games and devices can be adapted to enable
play, but for some players these adaptations may not go far enough. Games may require
inputs that some players cannot provide with any device. To address this problem,
we introduce partial automation, an accessibility technique that delegates control
of inaccessible game inputs to an AI partner. Partial automation complements and builds
on other approaches to improving games' accessibility, including universal design,
player balancing, and interface adaptation. We have demonstrated partial automation
in two games for the rehabilitation of spinal cord injury. Six study participants
with vastly different motor abilities were able to play both games. Participants liked
the increased personalization that partial automation affords, although some participants
were confused by aspects of the AI's behaviour.
Khaleghi, Leyla; Moghaddam, Alireza Sepas; Marshall, Joshua; Etemad, Ali
Multi-View Video-Based 3D Hand Pose Estimation Miscellaneous
2021.
Abstract | Links | BibTeX | Tags:
@misc{khaleghi2021multiview,
title = {Multi-View Video-Based 3D Hand Pose Estimation},
author = {Leyla Khaleghi and Alireza Sepas Moghaddam and Joshua Marshall and Ali Etemad},
url = {https://arxiv.org/pdf/2109.11747.pdf},
year = {2021},
date = {2021-09-24},
urldate = {2021-01-01},
abstract = {Hand pose estimation (HPE) can be used for a variety of human-computer interaction applications such as gesture-based control for physical or virtual/augmented reality devices. Recent works have shown that videos or multi-view images carry rich information regarding the hand, allowing for the development of more robust HPE systems. In this paper, we present the Multi-View Video-Based 3D Hand (MuViHand) dataset, consisting of multi-view videos of the hand along with ground-truth 3D pose labels. Our dataset includes more than 402,000 synthetic hand images available in 4,560 videos. The videos have been simultaneously captured from six different angles with complex backgrounds and random levels of dynamic lighting. The data has been captured from 10 distinct animated subjects using 12 cameras in a semi-circle topology where six tracking cameras only focus on the hand and the other six fixed cameras capture the entire body. Next, we implement MuViHandNet, a neural pipeline consisting of image encoders for obtaining visual embeddings of the hand, recurrent learners to learn both temporal and angular sequential information, and graph networks with U-Net architectures to estimate the final 3D pose information. We perform extensive experiments and show the challenging nature of this new dataset as well as the effectiveness of our proposed method. Ablation studies show the added value of each component in MuViHandNet, as well as the benefit of having temporal and sequential information in the dataset. },
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Hand pose estimation (HPE) can be used for a variety of human-computer interaction applications such as gesture-based control for physical or virtual/augmented reality devices. Recent works have shown that videos or multi-view images carry rich information regarding the hand, allowing for the development of more robust HPE systems. In this paper, we present the Multi-View Video-Based 3D Hand (MuViHand) dataset, consisting of multi-view videos of the hand along with ground-truth 3D pose labels. Our dataset includes more than 402,000 synthetic hand images available in 4,560 videos. The videos have been simultaneously captured from six different angles with complex backgrounds and random levels of dynamic lighting. The data has been captured from 10 distinct animated subjects using 12 cameras in a semi-circle topology where six tracking cameras only focus on the hand and the other six fixed cameras capture the entire body. Next, we implement MuViHandNet, a neural pipeline consisting of image encoders for obtaining visual embeddings of the hand, recurrent learners to learn both temporal and angular sequential information, and graph networks with U-Net architectures to estimate the final 3D pose information. We perform extensive experiments and show the challenging nature of this new dataset as well as the effectiveness of our proposed method. Ablation studies show the added value of each component in MuViHandNet, as well as the benefit of having temporal and sequential information in the dataset.
Zhang, Guangyi; Etemad, Ali
Holistic Semi-Supervised Approaches for EEG Representation Learning Miscellaneous
2021.
Abstract | Links | BibTeX | Tags:
@misc{zhang2021holistic,
title = {Holistic Semi-Supervised Approaches for EEG Representation Learning},
author = {Guangyi Zhang and Ali Etemad},
url = {https://arxiv.org/pdf/2109.11732.pdf},
year = {2021},
date = {2021-09-24},
urldate = {2021-01-01},
abstract = {Recently, supervised methods, which often require substantial amounts of class labels, have achieved promising results for EEG representation learning. However, labeling EEG data is a challenging task. More recently, holistic semi-supervised learning approaches, which only require few output labels, have shown promising results in the field of computer vision. These methods, however, have not yet been adapted for EEG learning. In this paper, we adapt three state-of-the-art holistic semi-supervised approaches, namely MixMatch, FixMatch, and AdaMatch, as well as five classical semi-supervised methods for EEG learning. We perform rigorous experiments with all 8 methods on two public EEG-based emotion recognition datasets, namely SEED and SEED-IV. The experiments with different amounts of limited labeled samples show that the holistic approaches achieve strong results even when only 1 labeled sample is used per class. Further experiments show that in most cases, AdaMatch is the most effective method, followed by MixMatch and FixMatch. },
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Recently, supervised methods, which often require substantial amounts of class labels, have achieved promising results for EEG representation learning. However, labeling EEG data is a challenging task. More recently, holistic semi-supervised learning approaches, which only require few output labels, have shown promising results in the field of computer vision. These methods, however, have not yet been adapted for EEG learning. In this paper, we adapt three state-of-the-art holistic semi-supervised approaches, namely MixMatch, FixMatch, and AdaMatch, as well as five classical semi-supervised methods for EEG learning. We perform rigorous experiments with all 8 methods on two public EEG-based emotion recognition datasets, namely SEED and SEED-IV. The experiments with different amounts of limited labeled samples show that the holistic approaches achieve strong results even when only 1 labeled sample is used per class. Further experiments show that in most cases, AdaMatch is the most effective method, followed by MixMatch and FixMatch.
Yang, Xiaoyu; Zhu, Xiaodan
Exploring Decomposition for Table-based Fact Verification Miscellaneous
2021.
BibTeX | Tags:
@misc{yang2021exploring,
title = {Exploring Decomposition for Table-based Fact Verification},
author = {Xiaoyu Yang and Xiaodan Zhu},
year = {2021},
date = {2021-09-22},
urldate = {2021-01-01},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Taghanaki, Setareh Rahimi; Rainbow, Michael J; Etemad, Ali
Self-Supervised Human Activity Recognition by Learning to Predict Cross-Dimensional Motion Inproceedings
In: 2021 International Symposium on Wearable Computers, pp. 23–27, Association for Computing Machinery, Virtual, USA, 2021, ISBN: 9781450384629.
Abstract | Links | BibTeX | Tags: Accelerometer Data, Data Prediction, Human Activity Recognition, Machine Learning., Self-supervised Learning
@inproceedings{10.1145/3460421.3480417,
title = {Self-Supervised Human Activity Recognition by Learning to Predict Cross-Dimensional Motion},
author = {Setareh Rahimi Taghanaki and Michael J Rainbow and Ali Etemad},
url = {https://doi.org/10.1145/3460421.3480417},
doi = {10.1145/3460421.3480417},
isbn = {9781450384629},
year = {2021},
date = {2021-09-21},
urldate = {2021-01-01},
booktitle = {2021 International Symposium on Wearable Computers},
pages = {23–27},
publisher = {Association for Computing Machinery},
address = {Virtual, USA},
series = {ISWC '21},
abstract = {We propose the use of self-supervised learning for human activity recognition with
smartphone accelerometer data. Our proposed solution consists of two steps. First,
the representations of unlabeled input signals are learned by training a deep convolutional
neural network to predict a segment of accelerometer values. Our model exploits a
novel scheme to leverage past and present motion in x and y dimensions, as well as
past values of the z axis to predict values in the z dimension. This cross-dimensional
prediction approach results in effective pretext training with which our model learns
to extract strong representations. Next, we freeze the convolution blocks and transfer
the weights to our downstream network aimed at human activity recognition. For this
task, we add a number of fully connected layers to the end of the frozen network and
train the added layers with labeled accelerometer signals to learn to classify human
activities. We evaluate the performance of our method on three publicly available
human activity datasets: UCI HAR, MotionSense, and HAPT. The results show that our
approach outperforms the existing methods and sets new state-of-the-art results.},
keywords = {Accelerometer Data, Data Prediction, Human Activity Recognition, Machine Learning., Self-supervised Learning},
pubstate = {published},
tppubtype = {inproceedings}
}
We propose the use of self-supervised learning for human activity recognition with
smartphone accelerometer data. Our proposed solution consists of two steps. First,
the representations of unlabeled input signals are learned by training a deep convolutional
neural network to predict a segment of accelerometer values. Our model exploits a
novel scheme to leverage past and present motion in x and y dimensions, as well as
past values of the z axis to predict values in the z dimension. This cross-dimensional
prediction approach results in effective pretext training with which our model learns
to extract strong representations. Next, we freeze the convolution blocks and transfer
the weights to our downstream network aimed at human activity recognition. For this
task, we add a number of fully connected layers to the end of the frozen network and
train the added layers with labeled accelerometer signals to learn to classify human
activities. We evaluate the performance of our method on three publicly available
human activity datasets: UCI HAR, MotionSense, and HAPT. The results show that our
approach outperforms the existing methods and sets new state-of-the-art results.
smartphone accelerometer data. Our proposed solution consists of two steps. First,
the representations of unlabeled input signals are learned by training a deep convolutional
neural network to predict a segment of accelerometer values. Our model exploits a
novel scheme to leverage past and present motion in x and y dimensions, as well as
past values of the z axis to predict values in the z dimension. This cross-dimensional
prediction approach results in effective pretext training with which our model learns
to extract strong representations. Next, we freeze the convolution blocks and transfer
the weights to our downstream network aimed at human activity recognition. For this
task, we add a number of fully connected layers to the end of the frozen network and
train the added layers with labeled accelerometer signals to learn to classify human
activities. We evaluate the performance of our method on three publicly available
human activity datasets: UCI HAR, MotionSense, and HAPT. The results show that our
approach outperforms the existing methods and sets new state-of-the-art results.
Caldwell, Jack; Marshall, Joshua A.
Towards efficient learning-based model predictive control via feedback linearization and Gaussian process regression Journal Article
In: IEEE, 2021.
@article{Caldwell2021,
title = {Towards efficient learning-based model predictive control via feedback linearization and Gaussian process regression},
author = {Jack Caldwell and Joshua A. Marshall},
url = {https://qspace.library.queensu.ca/handle/1974/29514},
year = {2021},
date = {2021-09-09},
urldate = {2021-09-09},
journal = {IEEE},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Yang, Xiaoyu; Zhu, Xiaodan; Shi, Zhan; Li, Tianda
Unsupervised Pre-training with Structured Knowledge for Improving Natural Language Inference Miscellaneous
2021.
Abstract | Links | BibTeX | Tags:
@misc{yang2021unsupervised,
title = {Unsupervised Pre-training with Structured Knowledge for Improving Natural Language Inference},
author = {Xiaoyu Yang and Xiaodan Zhu and Zhan Shi and Tianda Li},
url = {https://arxiv.org/pdf/2109.03941.pdf},
year = {2021},
date = {2021-09-08},
urldate = {2021-09-08},
abstract = {While recent research on natural language inference has considerably benefited from large annotated datasets, the amount of inference-related knowledge (including commonsense) provided in the annotated data is still rather limited. There have been two lines of approaches that can be used to further address the limitation: (1) unsupervised pretraining can leverage knowledge in much larger unstructured text data; (2) structured (often human-curated) knowledge has started to be considered in neural-network-based models for NLI. An immediate question is whether these two approaches complement each other, or how to develop models that can bring together their advantages. In this paper, we propose models that leverage structured knowledge in different components of pre-trained models. Our results show that the proposed models perform better than previous BERT-based state-of-the-art models. Although our models are proposed for NLI, they can be easily extended to other sentence or sentence-pair classification problems. },
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
While recent research on natural language inference has considerably benefited from large annotated datasets, the amount of inference-related knowledge (including commonsense) provided in the annotated data is still rather limited. There have been two lines of approaches that can be used to further address the limitation: (1) unsupervised pretraining can leverage knowledge in much larger unstructured text data; (2) structured (often human-curated) knowledge has started to be considered in neural-network-based models for NLI. An immediate question is whether these two approaches complement each other, or how to develop models that can bring together their advantages. In this paper, we propose models that leverage structured knowledge in different components of pre-trained models. Our results show that the proposed models perform better than previous BERT-based state-of-the-art models. Although our models are proposed for NLI, they can be easily extended to other sentence or sentence-pair classification problems.
Kourkounakis, Tedd; Hajavi, Amirhossein; Etemad, Ali
FluentNet: End-to-End Detection of Stuttered Speech Disfluencies with Deep Learning Journal Article
In: IEEE/ACM Transactions on Audio, Speech, and Language Processing, pp. 1-1, 2021.
@article{9528931,
title = {FluentNet: End-to-End Detection of Stuttered Speech Disfluencies with Deep Learning},
author = {Tedd Kourkounakis and Amirhossein Hajavi and Ali Etemad},
doi = {10.1109/TASLP.2021.3110146},
year = {2021},
date = {2021-09-03},
urldate = {2021-01-01},
journal = {IEEE/ACM Transactions on Audio, Speech, and Language Processing},
pages = {1-1},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Mokogwu, Chiedu N.; Hashtrudi–Zaad, Keyvan
Energy-Based Analysis of String Stability in Heterogeneous Platoons Inproceedings
In: 2021 IEEE 94th Vehicular Technology Conference (VTC2021-Fall), pp. 01-06, 2021, ISSN: 2577-2465.
Abstract | Links | BibTeX | Tags:
@inproceedings{9625178,
title = {Energy-Based Analysis of String Stability in Heterogeneous Platoons},
author = {Chiedu N. Mokogwu and Keyvan Hashtrudi–Zaad},
url = {https://ieeexplore.ieee.org/document/9625178},
doi = {10.1109/VTC2021-Fall52928.2021.9625178},
issn = {2577-2465},
year = {2021},
date = {2021-09-01},
urldate = {2021-09-01},
booktitle = {2021 IEEE 94th Vehicular Technology Conference (VTC2021-Fall)},
pages = {01-06},
abstract = {Analysis of string stability in vehicle platoons has mainly been confined to strings of vehicles having similar dynamics. In this scenario, errors propagate uniformly downstream in the string, making string stability analysis convenient. In real driving and platooning scenarios, homogeneity of vehicles is highly unlikely. This means errors amplify/attenuate in a nonuniform manner downstream, as a result of differing dynamics of the current and preceding vehicles appearing in the platoon error dynamic model. This complicates string stability analysis. To remedy this issue, a dynamic proxy to decouple the error dynamics of current vehicle from its predecessor is proposed. Then, by utilizing passivity-based techniques to constrain the output strict passivity index of each vehicle subsystem in the string to unity through a two-stage control strategy, the condition for string stability is met. Validation of the proposed strategy through numerical simulations is nresented.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Analysis of string stability in vehicle platoons has mainly been confined to strings of vehicles having similar dynamics. In this scenario, errors propagate uniformly downstream in the string, making string stability analysis convenient. In real driving and platooning scenarios, homogeneity of vehicles is highly unlikely. This means errors amplify/attenuate in a nonuniform manner downstream, as a result of differing dynamics of the current and preceding vehicles appearing in the platoon error dynamic model. This complicates string stability analysis. To remedy this issue, a dynamic proxy to decouple the error dynamics of current vehicle from its predecessor is proposed. Then, by utilizing passivity-based techniques to constrain the output strict passivity index of each vehicle subsystem in the string to unity through a two-stage control strategy, the condition for string stability is met. Validation of the proposed strategy through numerical simulations is nresented.
Hajian, Gelareh; Etemad, Ali; Morin, Evelyn
Generalized EMG-based isometric contact force estimation using a deep learning approach Journal Article
In: Biomedical Signal Processing and Control, vol. 70, pp. 103012, 2021, ISSN: 1746-8094.
Abstract | Links | BibTeX | Tags: Convolutional neural network (CNN), Force estimation, High-density (HD) recording, Isometric contraction, Surface electromyogram (EMG)
@article{HAJIAN2021103012,
title = {Generalized EMG-based isometric contact force estimation using a deep learning approach},
author = {Gelareh Hajian and Ali Etemad and Evelyn Morin},
url = {https://www.sciencedirect.com/science/article/pii/S1746809421006091},
doi = {https://doi.org/10.1016/j.bspc.2021.103012},
issn = {1746-8094},
year = {2021},
date = {2021-09-01},
journal = {Biomedical Signal Processing and Control},
volume = {70},
pages = {103012},
abstract = {EMG-based force estimation is generally done in a subject specific manner. In this paper, we explore force estimation in a manner generalizable across individuals, where the EMG signals are recorded from the long head and the short head of biceps brachii, and the brachioradialis, under isometric elbow flexion, while the contact force is measured at the wrist. Deep convolutional neural networks (CNN), which utilize feature-level fusion of representations learned from high-density (HD) EMG in the time and frequency domains, are developed. The performance of the proposed solution (CNN-FLF) is compared to a number of baselines including CNNs with input-level fusion of HD-EMG data in the time and frequency domains, CNNs which estimate force from time or frequency domain EMG data separately, and a number of classical machine learning methods, which use hand-crafted features extracted from the EMG signals. Results show that the CNN-FLF, with optimized hyper-parameters, outperformed all other methods, giving a normalized mean squared error for estimated force of 1.6±3.69% (mean±SD). In visualization of the extracted features for the different CNN models, it is apparent that the final features of the CNN-FLF enable finding an accurate regressive relationship with output force levels.},
keywords = {Convolutional neural network (CNN), Force estimation, High-density (HD) recording, Isometric contraction, Surface electromyogram (EMG)},
pubstate = {published},
tppubtype = {article}
}
EMG-based force estimation is generally done in a subject specific manner. In this paper, we explore force estimation in a manner generalizable across individuals, where the EMG signals are recorded from the long head and the short head of biceps brachii, and the brachioradialis, under isometric elbow flexion, while the contact force is measured at the wrist. Deep convolutional neural networks (CNN), which utilize feature-level fusion of representations learned from high-density (HD) EMG in the time and frequency domains, are developed. The performance of the proposed solution (CNN-FLF) is compared to a number of baselines including CNNs with input-level fusion of HD-EMG data in the time and frequency domains, CNNs which estimate force from time or frequency domain EMG data separately, and a number of classical machine learning methods, which use hand-crafted features extracted from the EMG signals. Results show that the CNN-FLF, with optimized hyper-parameters, outperformed all other methods, giving a normalized mean squared error for estimated force of 1.6±3.69% (mean±SD). In visualization of the extracted features for the different CNN models, it is apparent that the final features of the CNN-FLF enable finding an accurate regressive relationship with output force levels.
Taghanaki, Setareh Rahimi; Rainbow, Michael; Etemad, Ali
Wearable-based Classification of Running Styles with Deep Learning Miscellaneous
2021.
BibTeX | Tags:
@misc{taghanaki2021wearablebased,
title = {Wearable-based Classification of Running Styles with Deep Learning},
author = {Setareh Rahimi Taghanaki and Michael Rainbow and Ali Etemad},
year = {2021},
date = {2021-09-01},
urldate = {2021-01-01},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Taghanaki, Setareh Rahimi; Rainbow, Michael; Etemad, Ali
Wearable-based Classification of Running Styles with Deep Learning Journal Article
In: arXiv e-prints, pp. arXiv:2109.00594, 2021.
BibTeX | Tags: Computer Science - Machine Learning, Electrical Engineering and Systems Science - Signal Processing
@article{2021arXiv210900594R,
title = {Wearable-based Classification of Running Styles with Deep Learning},
author = {Setareh Rahimi Taghanaki and Michael Rainbow and Ali Etemad},
year = {2021},
date = {2021-09-01},
urldate = {2021-09-01},
journal = {arXiv e-prints},
pages = {arXiv:2109.00594},
keywords = {Computer Science - Machine Learning, Electrical Engineering and Systems Science - Signal Processing},
pubstate = {published},
tppubtype = {article}
}
Lee, Sunghoon; Park, Jooyoun; Kim, Il-Min; Heo, Jun
Flexible soft-output decoding of polar codes Journal Article
In: EURASIP Journal on Wireless Communications and Networking, vol. 2021, 2021, ISBN: 1687-1499.
Abstract | Links | BibTeX | Tags:
@article{Lee2021,
title = {Flexible soft-output decoding of polar codes},
author = { Sunghoon Lee and Jooyoun Park and Il-Min Kim and Jun Heo},
url = {https://jwcn-eurasipjournals.springeropen.com/articles/10.1186/s13638-021-02042-x#Abs1},
doi = {10.1186/s13638-021-02042-x},
isbn = {1687-1499},
year = {2021},
date = {2021-08-24},
urldate = {2021-08-24},
journal = {EURASIP Journal on Wireless Communications and Networking},
volume = {2021},
abstract = { In this research, we study soft-output decoding of polar codes. Two representative soft-output decoding algorithms are belief propagation (BP) and soft cancellation (SCAN). The BP algorithm has low latency but suffers from high computational complexity. On the other hand, the SCAN algorithm, which is proposed for reduced complexity of soft-output decoding, achieves good decoding performance but suffers from long latency. These two algorithms are suitable only for two extreme cases that need very low latency (but with high complexity) or very low complexity (but with high latency). However, many practical systems may need to work for the moderate cases (i.e., not too high latency and not too high complexity) rather than two extremes. To adapt to the various needs of the systems, we propose a very flexible soft-output decoding framework of polar codes. Depending on which system requirement is most crucial, the proposed scheme can adapt to the systems by controlling the level of parallelism. Numerical results demonstrate that the proposed scheme can effectively adapt to various system requirements by changing the level of parallelism.},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
In this research, we study soft-output decoding of polar codes. Two representative soft-output decoding algorithms are belief propagation (BP) and soft cancellation (SCAN). The BP algorithm has low latency but suffers from high computational complexity. On the other hand, the SCAN algorithm, which is proposed for reduced complexity of soft-output decoding, achieves good decoding performance but suffers from long latency. These two algorithms are suitable only for two extreme cases that need very low latency (but with high complexity) or very low complexity (but with high latency). However, many practical systems may need to work for the moderate cases (i.e., not too high latency and not too high complexity) rather than two extremes. To adapt to the various needs of the systems, we propose a very flexible soft-output decoding framework of polar codes. Depending on which system requirement is most crucial, the proposed scheme can adapt to the systems by controlling the level of parallelism. Numerical results demonstrate that the proposed scheme can effectively adapt to various system requirements by changing the level of parallelism.
Forbrigger, Shane; Hashtrudi-Zaad, Keyvan
An Intent-Preserving Approach to Telerobotic Rehabilitation Inproceedings
In: 2021 IEEE World Haptics Conference (WHC), pp. 739-744, 2021.
@inproceedings{9517196,
title = {An Intent-Preserving Approach to Telerobotic Rehabilitation},
author = {Shane Forbrigger and Keyvan Hashtrudi-Zaad},
doi = {10.1109/WHC49131.2021.9517196},
year = {2021},
date = {2021-08-23},
urldate = {2021-01-01},
booktitle = {2021 IEEE World Haptics Conference (WHC)},
pages = {739-744},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Pecly, Leonam; Hashtrudi-Zaad, Keyvan
Analog Position using Sensor Fusion for Haptic Systems Control Inproceedings
In: 2021 IEEE World Haptics Conference (WHC), pp. 391-396, 2021.
@inproceedings{9517229,
title = {Analog Position using Sensor Fusion for Haptic Systems Control},
author = {Leonam Pecly and Keyvan Hashtrudi-Zaad},
doi = {10.1109/WHC49131.2021.9517229},
year = {2021},
date = {2021-08-23},
urldate = {2021-01-01},
booktitle = {2021 IEEE World Haptics Conference (WHC)},
pages = {391-396},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Pecly, Leonam; Hashtrudi-Zaad, Keyvan
Uncoupled Stability Region for Analog and Discrete Implementation of Mass-Damper-Spring Virtual Environments Inproceedings
In: 2021 IEEE World Haptics Conference (WHC), pp. 864-864, 2021.
@inproceedings{9517241,
title = {Uncoupled Stability Region for Analog and Discrete Implementation of Mass-Damper-Spring Virtual Environments},
author = {Leonam Pecly and Keyvan Hashtrudi-Zaad},
doi = {10.1109/WHC49131.2021.9517241},
year = {2021},
date = {2021-08-23},
urldate = {2021-01-01},
booktitle = {2021 IEEE World Haptics Conference (WHC)},
pages = {864-864},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Pecly, Leonam; Hashtrudi-Zaad, Keyvan
Enhanced Haptic Passivity and Uncoupled Stability with Tustin-like General Discretization Method Inproceedings
In: 2021 IEEE World Haptics Conference (WHC), pp. 109-114, 2021.
@inproceedings{9517253,
title = {Enhanced Haptic Passivity and Uncoupled Stability with Tustin-like General Discretization Method},
author = {Leonam Pecly and Keyvan Hashtrudi-Zaad},
doi = {10.1109/WHC49131.2021.9517253},
year = {2021},
date = {2021-08-23},
urldate = {2021-01-01},
booktitle = {2021 IEEE World Haptics Conference (WHC)},
pages = {109-114},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Ebrahimi, Javad; Salari, Omid; Eren, Suzan; Zaad, Keyvan Hashtrudi; Bakhshai, Alireza; Jain, Praveen
Efficiency Improved Multi-Source Inverter for Hybrid Energy Storage Systems in Electric Vehicle Application Journal Article
In: IEEE Transactions on Power Electronics, pp. 1-1, 2021.
@article{9513596,
title = {Efficiency Improved Multi-Source Inverter for Hybrid Energy Storage Systems in Electric Vehicle Application},
author = {Javad Ebrahimi and Omid Salari and Suzan Eren and Keyvan Hashtrudi Zaad and Alireza Bakhshai and Praveen Jain},
doi = {10.1109/TPEL.2021.3104759},
year = {2021},
date = {2021-08-13},
journal = {IEEE Transactions on Power Electronics},
pages = {1-1},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Simon, Nisha; Muise, Christian
A Natural Language Model for Generating PDDL Journal Article
In: 2021.
@article{simonnatural,
title = {A Natural Language Model for Generating PDDL},
author = {Nisha Simon and Christian Muise},
url = {https://icaps21.icaps-conference.org/workshops/KEPS/Papers/KEPS_2021_paper_7.pdf},
year = {2021},
date = {2021-08-12},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Kearns, Oliver; Moore, Ian D; Hoult, Neil A
Experimental Study of a Noncircular Corrugated Steel Culvert at Different Shallow Cover Depths Book Chapter
In: Pipelines 2021, pp. 188-198, 2021.
Abstract | Links | BibTeX | Tags:
@inbook{doi:10.1061/9780784483619.020,
title = {Experimental Study of a Noncircular Corrugated Steel Culvert at Different Shallow Cover Depths},
author = {Oliver Kearns and Ian D Moore and Neil A Hoult},
url = {https://ascelibrary.org/doi/abs/10.1061/9780784483619.020},
doi = {10.1061/9780784483619.020},
year = {2021},
date = {2021-08-12},
booktitle = {Pipelines 2021},
pages = {188-198},
abstract = {North American bridge design codes recommend a minimum burial depth for culverts to prevent failure during construction or under vehicle loading. The design of flexible culverts, such as those made of corrugated steel plate, has historically focused on circumferential force in the conduit walls (i.e., hoop thrust) as the critical failure mode. However, recent studies have shown that bending stresses are often of the same magnitude as hoop stresses, especially for noncircular pipe culverts with cover depths that are at or near the recommended minimum level. To investigate this issue, an elliptical corrugated steel culvert with span of 1,650 mm, rise of 1,300 mm, and corrugation profile with a 25 mm amplitude and 75 mm wavelength was tested at different cover depths ranging from 0.1 to 1.2 m under simulated vehicle loading. The culvert’s response to live load quantifies just how much more dominant bending moment is than thrust. When combined earth and live loading is considered, moments are still important, and the results quantify the transition from a flexural, live load dominated response at extremely shallow cover depths to a more earth load dominated response at cover depths greater than 500 mm.},
keywords = {},
pubstate = {published},
tppubtype = {inbook}
}
North American bridge design codes recommend a minimum burial depth for culverts to prevent failure during construction or under vehicle loading. The design of flexible culverts, such as those made of corrugated steel plate, has historically focused on circumferential force in the conduit walls (i.e., hoop thrust) as the critical failure mode. However, recent studies have shown that bending stresses are often of the same magnitude as hoop stresses, especially for noncircular pipe culverts with cover depths that are at or near the recommended minimum level. To investigate this issue, an elliptical corrugated steel culvert with span of 1,650 mm, rise of 1,300 mm, and corrugation profile with a 25 mm amplitude and 75 mm wavelength was tested at different cover depths ranging from 0.1 to 1.2 m under simulated vehicle loading. The culvert’s response to live load quantifies just how much more dominant bending moment is than thrust. When combined earth and live loading is considered, moments are still important, and the results quantify the transition from a flexural, live load dominated response at extremely shallow cover depths to a more earth load dominated response at cover depths greater than 500 mm.
Tripakis, Stavros; Rudie, Karen
Decentralized Observation of Discrete-Event Systems: At Least One Can Tell Miscellaneous
2021.
BibTeX | Tags:
@misc{tripakis2021decentralized,
title = {Decentralized Observation of Discrete-Event Systems: At Least One Can Tell},
author = {Stavros Tripakis and Karen Rudie},
year = {2021},
date = {2021-08-10},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Showalter, Ian; Schwartz, Howard; Givigi, Sidney N.
Evolutionary Inherited Neuromodulated Neurocontrollers with Objective Weighted Ranking Inproceedings
In: 2021 IEEE Congress on Evolutionary Computation (CEC), pp. 2443-2450, 2021.
@inproceedings{9504892,
title = {Evolutionary Inherited Neuromodulated Neurocontrollers with Objective Weighted Ranking},
author = {Ian Showalter and Howard Schwartz and Sidney N. Givigi},
doi = {10.1109/CEC45853.2021.9504892},
year = {2021},
date = {2021-08-09},
booktitle = {2021 IEEE Congress on Evolutionary Computation (CEC)},
pages = {2443-2450},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Roy, Shuvendu; Etemad, Ali
Spatiotemporal Contrastive Learning of Facial Expressions in Videos Miscellaneous
2021.
BibTeX | Tags:
@misc{roy2021spatiotemporal,
title = {Spatiotemporal Contrastive Learning of Facial Expressions in Videos},
author = {Shuvendu Roy and Ali Etemad},
year = {2021},
date = {2021-08-06},
urldate = {2021-08-06},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Bhatti, Anubhav; Behinaein, Behnam; Rodenburg, Dirk; Hungler, Paul; Etemad, Ali
Attentive Cross-modal Connections for Deep Multimodal Wearable-based Emotion Recognition Miscellaneous
2021.
@misc{bhatti2021attentive,
title = {Attentive Cross-modal Connections for Deep Multimodal Wearable-based Emotion Recognition},
author = {Anubhav Bhatti and Behnam Behinaein and Dirk Rodenburg and Paul Hungler and Ali Etemad},
url = {https://arxiv.org/pdf/2108.02241.pdf},
year = {2021},
date = {2021-08-04},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Zhou, Yuxuan; Zhou, Kaiyin; Liu, Xien; Wu, Ji; Zhu, Xiaodan
THiFly_Queens at SemEval-2021 Task 9: Two-stage Statement Verification with Adaptive Ensembling and Slot-based Operation Inproceedings
In: Proceedings of the 15th International Workshop on Semantic Evaluation (SemEval-2021), pp. 416–422, Ässociation for Computational Linguistics, Önline, 2021.
Abstract | Links | BibTeX | Tags:
@inproceedings{zhou-etal-2021-thifly,
title = {THiFly_Queens at SemEval-2021 Task 9: Two-stage Statement Verification with Adaptive Ensembling and Slot-based Operation},
author = {Yuxuan Zhou and Kaiyin Zhou and Xien Liu and Ji Wu and Xiaodan Zhu},
url = {https://aclanthology.org/2021.semeval-1.50},
doi = {10.18653/v1/2021.semeval-1.50},
year = {2021},
date = {2021-08-01},
urldate = {2021-08-01},
booktitle = {Proceedings of the 15th International Workshop on Semantic Evaluation (SemEval-2021)},
pages = {416--422},
publisher = {Ässociation for Computational Linguistics},
address = {Önline},
abstract = {This paper describes our system for verifying statements with tables at SemEval-2021 Task 9. We developed a two-stage verifying system based on the latest table-based pre-trained model GraPPa. Multiple networks are devised to verify different types of statements in the competition dataset and an adaptive model ensembling technique is applied to ensemble models in both stages. A statement-slot-based symbolic operation module is also used in our system to further improve the performance and stability of the system. Our model achieves second place in the 3-way classification and fourth place in the 2-way classification evaluation. Several ablation experiments show the effectiveness of different modules proposed in this paper.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
This paper describes our system for verifying statements with tables at SemEval-2021 Task 9. We developed a two-stage verifying system based on the latest table-based pre-trained model GraPPa. Multiple networks are devised to verify different types of statements in the competition dataset and an adaptive model ensembling technique is applied to ensemble models in both stages. A statement-slot-based symbolic operation module is also used in our system to further improve the performance and stability of the system. Our model achieves second place in the 3-way classification and fourth place in the 2-way classification evaluation. Several ablation experiments show the effectiveness of different modules proposed in this paper.
Roy, Jeremy
Autonomous Ground Vehicle Guidance and Mapping in Challenging Terrains by using On-Board Vibration Measurements Journal Article
In: IEEE, 2021.
BibTeX | Tags:
@article{Roy2021,
title = {Autonomous Ground Vehicle Guidance and Mapping in Challenging Terrains by using On-Board Vibration Measurements},
author = {Jeremy Roy},
year = {2021},
date = {2021-07-29},
journal = {IEEE},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Zhang, Guangyi; Etemad, Ali
Deep Recurrent Semi-Supervised EEG Representation Learning for Emotion Recognition Miscellaneous
2021.
@misc{zhang2021deep,
title = {Deep Recurrent Semi-Supervised EEG Representation Learning for Emotion Recognition},
author = {Guangyi Zhang and Ali Etemad},
url = {https://arxiv.org/pdf/2107.13505.pdf},
year = {2021},
date = {2021-07-28},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Ean, Richard; Rudie, Karen
Do What You Know: Coupling Knowledge with Action in Discrete-Event Systems Miscellaneous
2021.
@misc{ean2021know,
title = {Do What You Know: Coupling Knowledge with Action in Discrete-Event Systems},
author = {Richard Ean and Karen Rudie},
url = {https://arxiv.org/pdf/2108.02000.pdf
},
year = {2021},
date = {2021-07-25},
keywords = {},
pubstate = {published},
tppubtype = {misc}
}
Fuzheng; Hoult Sun, Neil; Butler
Distributed monitoring of rail lateral buckling under axial loading Journal Article
In: Journal of Civil Structural Health Monitoring, 2021.
Abstract | Links | BibTeX | Tags: Fibre Optic Sensors, Rail monitoring
@article{Sun2021,
title = {Distributed monitoring of rail lateral buckling under axial loading},
author = {Sun, Fuzheng; Hoult, Neil; Butler, Liam J.; Zhang, Merrina},
url = {https://link.springer.com/article/10.1007/s13349-021-00504-w},
doi = {https://doi.org/10.1007/s13349-021-00504-w},
year = {2021},
date = {2021-07-15},
journal = {Journal of Civil Structural Health Monitoring},
abstract = {Distributed fiber optic sensors (DFOS) were used, for the first time, to monitor lateral buckling of a rail under axial loading with different boundary conditions. In the experiments, two types of DFOS systems were used to collect the distributed strain data, and the performance of the two systems was compared. The distributed strain data were used to develop a fitted strain plane at each cross-section along the rail, and axial strain and bending curvature were derived from the fitted strain plane. A data extrapolation method using the distributed curvature data was developed to evaluate the actual boundary conditions at the ends of the rail since they did not match assumed ideal behaviour. The distributed rail deflection along the length of rail was calculated by integrating distributed curvature data and the results were compared with the deflection measured by linear potentiometers (LPs).},
keywords = {Fibre Optic Sensors, Rail monitoring},
pubstate = {published},
tppubtype = {article}
}
Distributed fiber optic sensors (DFOS) were used, for the first time, to monitor lateral buckling of a rail under axial loading with different boundary conditions. In the experiments, two types of DFOS systems were used to collect the distributed strain data, and the performance of the two systems was compared. The distributed strain data were used to develop a fitted strain plane at each cross-section along the rail, and axial strain and bending curvature were derived from the fitted strain plane. A data extrapolation method using the distributed curvature data was developed to evaluate the actual boundary conditions at the ends of the rail since they did not match assumed ideal behaviour. The distributed rail deflection along the length of rail was calculated by integrating distributed curvature data and the results were compared with the deflection measured by linear potentiometers (LPs).
Mayuku, Orighomisan; Surgenor, Brian W; Marshall, Joshua A
Multi-Resolution and Multi-Domain Analysis of Off-Road Datasets for Autonomous Driving Inproceedings
In: 2021 18th Conference on Robots and Vision (CRV), pp. 165-172, 2021.
@inproceedings{9469485,
title = {Multi-Resolution and Multi-Domain Analysis of Off-Road Datasets for Autonomous Driving},
author = {Orighomisan Mayuku and Brian W Surgenor and Joshua A Marshall},
doi = {10.1109/CRV52889.2021.00030},
year = {2021},
date = {2021-07-05},
booktitle = {2021 18th Conference on Robots and Vision (CRV)},
pages = {165-172},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Lins, Romulo Gonçalves; Givigi, Sidney N
In: IEEE Access, pp. 1-1, 2021.
@article{9475965,
title = {Cooperative Robotics and Machine Learning for Smart Manufacturing: Platform Design and Trends Within the Context of Industrial Internet of Things},
author = {Romulo Gonçalves Lins and Sidney N Givigi},
doi = {10.1109/ACCESS.2021.3094374},
year = {2021},
date = {2021-07-05},
journal = {IEEE Access},
pages = {1-1},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
Forbrigger, Shane; Hashtrudi-Zaad, Keyvan
An Intent-Preserving Approach to Telerobotic Rehabilitation Inproceedings
In: 2021 IEEE World Haptics Conference (WHC), pp. 739-744, 2021.
Abstract | Links | BibTeX | Tags:
@inproceedings{9517196b,
title = {An Intent-Preserving Approach to Telerobotic Rehabilitation},
author = {Shane Forbrigger and Keyvan Hashtrudi-Zaad},
url = {https://ieeexplore.ieee.org/document/9517196},
doi = {10.1109/WHC49131.2021.9517196},
year = {2021},
date = {2021-07-01},
urldate = {2021-07-01},
booktitle = {2021 IEEE World Haptics Conference (WHC)},
pages = {739-744},
abstract = {Often the design of control systems for telerobotic rehabilitation focuses on maintaining stability and maximizing transparency, but does not consider the intent of therapeutic interactions. When therapists assist patients during therapy activities, they tailor the amount of assistance and guidance delivered so that the patient engages fully in the therapy. Existing telerobotic control systems deliver the delayed therapist force to the patient, making only minimal changes as necessary to maintain stability. However, communication time delays can distort the effect of the therapist force and lead to remote interactions that do not reflect the therapist’s intent, such as over- or under-assisting the patient. In this work, we propose a method for identifying the therapist’s intent from their applied force and the velocity of the patient. Using this understanding of intent, we propose two approaches to ensure that the therapist’s intent is preserved across the communication channel: Rotational Intent-Preserving Teleoperation (RIPT), where the delayed force from the therapist is rotated to maintain the intended amount of assistance and guidance, and scaled-RIPT, where less relevant forces are reduced. We test these approaches in simulations, finding that they can prevent unintended over-assistance from the therapist.},
keywords = {},
pubstate = {published},
tppubtype = {inproceedings}
}
Often the design of control systems for telerobotic rehabilitation focuses on maintaining stability and maximizing transparency, but does not consider the intent of therapeutic interactions. When therapists assist patients during therapy activities, they tailor the amount of assistance and guidance delivered so that the patient engages fully in the therapy. Existing telerobotic control systems deliver the delayed therapist force to the patient, making only minimal changes as necessary to maintain stability. However, communication time delays can distort the effect of the therapist force and lead to remote interactions that do not reflect the therapist’s intent, such as over- or under-assisting the patient. In this work, we propose a method for identifying the therapist’s intent from their applied force and the velocity of the patient. Using this understanding of intent, we propose two approaches to ensure that the therapist’s intent is preserved across the communication channel: Rotational Intent-Preserving Teleoperation (RIPT), where the delayed force from the therapist is rotated to maintain the intended amount of assistance and guidance, and scaled-RIPT, where less relevant forces are reduced. We test these approaches in simulations, finding that they can prevent unintended over-assistance from the therapist.
Mokogwu, Chiedu N; Hashtrudi-Zaad, Keyvan
A hybrid position–rate teleoperation system Journal Article
In: Robotics and Autonomous Systems, vol. 141, pp. 103781, 2021, ISSN: 0921-8890.
Abstract | Links | BibTeX | Tags: Bilateral teleoperation, Position mode, Range sensor, Rate mode, Stability, Workspace expansion
@article{MOKOGWU2021103781,
title = {A hybrid position–rate teleoperation system},
author = {Chiedu N Mokogwu and Keyvan Hashtrudi-Zaad},
url = {https://www.sciencedirect.com/science/article/pii/S092188902100066X},
doi = {https://doi.org/10.1016/j.robot.2021.103781},
issn = {0921-8890},
year = {2021},
date = {2021-07-01},
journal = {Robotics and Autonomous Systems},
volume = {141},
pages = {103781},
abstract = {Position resolution is a major problem in teleoperation applications with significant disparity between master and slave workspace. While rate mode control is suitable for slave free motion operations, it poses significant stability and performance challenges for task manipulation. In this paper, we propose a hybrid control scheme that offers seamless transition between position and rate control modes based on the environment location information collected from a range sensor. The system incorporates the strengths of position and rate control modes while masking their shortcomings. Experiments to determine the viability of this method are carried out on a single degree-of-freedom teleoperation test-bed.},
keywords = {Bilateral teleoperation, Position mode, Range sensor, Rate mode, Stability, Workspace expansion},
pubstate = {published},
tppubtype = {article}
}
Position resolution is a major problem in teleoperation applications with significant disparity between master and slave workspace. While rate mode control is suitable for slave free motion operations, it poses significant stability and performance challenges for task manipulation. In this paper, we propose a hybrid control scheme that offers seamless transition between position and rate control modes based on the environment location information collected from a range sensor. The system incorporates the strengths of position and rate control modes while masking their shortcomings. Experiments to determine the viability of this method are carried out on a single degree-of-freedom teleoperation test-bed.