Ye Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Ye Wang has a total of 102 co-authors. Here are the top co-authors by number of shared publications:
- Toshiaki Koike-Akino
Mitsubishi Electric Research Laboratories (MERL), Cambridge, MA, USA
26 shared publications - Jing Liu
Mitsubishi Electric Research Laboratories (MERL)
13 shared publications Toshiaki Koike-Akino
Mitsubishi Electric Research Laboratories (MERL) 02139 Cambridge MA USA
13 shared publications- Kieran Parsons
Mitsubishi Electric Research Laboratories, Cambridge, MA, USA
10 shared publications - Deniz Erdogmus
Northeastern University, Boston, MA, USA
8 shared publications - Keisuke Kojima
Mitsubishi Electric Research Laboratories (MERL), 201 Broadway, Cambridge, MA 02139, USA
5 shared publications
Latest Publications
- Open Access
- 28 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 70724-70736
GPT Sonography: Hand Gesture Decoding From Forearm Ultrasound Images via a Large Vision-Language Model
Large vision-language models (LVLMs), such as the Generative Pre-trained Transformer 4-omni (GPT-4o), are emerging multi-modal foundation models which have great potential as powerful artificial-intelligence (AI) assistance tools for a myriad of applications, including healthcare, industrial, and academic sectors. Although such foundation models perform well in a wide range of general tasks, their capability without fine-tuning is often limited in specialized tasks. However, full fine-tuning of large foundation models is challenging due to enormous computation/memory/dataset requirements. Ultrasound data from the forearm has been shown to be used for hand gesture estimation. However, this typically requires training deep learning models with a large quantity of labeled data. We show that GPT-4o can decode hand gestures from forearm ultrasound data even with no fine-tuning, and improves with few-shot, retrieval augmented in-context learning. In our experiments, the average classification accuracy improved from 19.3% (0-shot) to 74.0% (2-shot) for within-session testing, and from 20.0% (0-shot) to 61.3%(3-shot) for cross-session testing. This demonstrates the potential of LVLMs for ultrasound-based gesture recognition by enabling an alternative to prior ultrasound gesture pipelines that require dedicated model training and large labeled datasets. Additionally, we show that few-shot in-context learning and retrieval-augmented selection can substantially improve performance without any model fine-tuning.
- 28 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 70724-70736
Open AccessGPT Sonography: Hand Gesture Decoding From Forearm Ultrasound Images via a Large Vision-Language Model
Large vision-language models (LVLMs), such as the Generative Pre-trained Transformer 4-omni (GPT-4o), are emerging multi-modal foundation models which have great potential as powerful artificial-intelligence (AI) assistance tools for a myriad of applications, including healthcare, industrial, and academic sectors. Although such foundation models perform well in a wide range of general tasks, their capability without fine-tuning is often limited in specialized tasks. However, full fine-tuning of large foundation models is challenging due to enormous computation/memory/dataset requirements. Ultrasound data from the forearm has been shown to be used for hand gesture estimation. However, this typically requires training deep learning models with a large quantity of labeled data. We show that GPT-4o can decode hand gestures from forearm ultrasound data even with no fine-tuning, and improves with few-shot, retrieval augmented in-context learning. In our experiments, the average classification accuracy improved from 19.3% (0-shot) to 74.0% (2-shot) for within-session testing, and from 20.0% (0-shot) to 61.3%(3-shot) for cross-session testing. This demonstrates the potential of LVLMs for ultrasound-based gesture recognition by enabling an alternative to prior ultrasound gesture pipelines that require dedicated model training and large labeled datasets. Additionally, we show that few-shot in-context learning and retrieval-augmented selection can substantially improve performance without any model fine-tuning.
- 6 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 22 (5) , 3634-3645
Temporal Surrogate Lagrangian Decomposition for Operational Hosting Capacity Assessment in Unbalanced Power Distribution Systems
The increasing penetration of distributed energy resources (DERs) necessitates advanced hosting capacity (HC) assessments to ensure reliable grid operation. Traditional static HC methods and dynamic operating envelopes often rely on instantaneous voltage constraints, leading to overly conservative estimates. In this article, we first propose a deterministic hosting capacity optimization model with a novel voltage violation duration constraint. Then, we build a graph convolutional recurrent network-based stochastic hosting capacity optimization framework. Finally, we propose a surrogate Lagrangian relaxation-based temporal decomposition scheme to improve the solving efficiency. Numerical results demonstrate the framework’s effectiveness in enhancing operational HC assessment and maximizing DER integration.
- 6 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. 22 (5) , 3634-3645
Temporal Surrogate Lagrangian Decomposition for Operational Hosting Capacity Assessment in Unbalanced Power Distribution Systems
The increasing penetration of distributed energy resources (DERs) necessitates advanced hosting capacity (HC) assessments to ensure reliable grid operation. Traditional static HC methods and dynamic operating envelopes often rely on instantaneous voltage constraints, leading to overly conservative estimates. In this article, we first propose a deterministic hosting capacity optimization model with a novel voltage violation duration constraint. Then, we build a graph convolutional recurrent network-based stochastic hosting capacity optimization framework. Finally, we propose a surrogate Lagrangian relaxation-based temporal decomposition scheme to improve the solving efficiency. Numerical results demonstrate the framework’s effectiveness in enhancing operational HC assessment and maximizing DER integration.
- 30 January 2026
- Published by Springer Nature
- p. 46-59
Quantum Diffusion Models for Few-Shot Learning
- 22 September 2025
- Published by Elsevier in Energy and Buildings
- Vol. 348
Probabilistic forecasting for building energy systems using time-series foundation models
- 22 September 2025
- Published by Elsevier in Energy and Buildings
- Vol. 348
Probabilistic forecasting for building energy systems using time-series foundation models
- Open Access
TuneComp: Joint Fine-tuning and Compression for Large Foundation Models
To reduce model size during post-training, compression methods, including knowledge distillation, low-rank approximation, and pruning, are often applied after fine-tuning the model. However, sequential fine-tuning and compression sacrifices performance, while creating a larger than necessary model as an intermediate step. In this work, we aim to reduce this gap, by directly constructing a smaller model while guided by the downstream task. We propose to jointly fine-tune and compress the model by gradually distilling it to a pruned low-rank structure. Experiments demonstrate that joint fine-tuning and compression significantly outperforms other sequential compression methods.
Open AccessTuneComp: Joint Fine-tuning and Compression for Large Foundation Models
To reduce model size during post-training, compression methods, including knowledge distillation, low-rank approximation, and pruning, are often applied after fine-tuning the model. However, sequential fine-tuning and compression sacrifices performance, while creating a larger than necessary model as an intermediate step. In this work, we aim to reduce this gap, by directly constructing a smaller model while guided by the downstream task. We propose to jointly fine-tune and compress the model by gradually distilling it to a pruned low-rank structure. Experiments demonstrate that joint fine-tuning and compression significantly outperforms other sequential compression methods.
- Open Access
LatentLLM: Attention-Aware Joint Tensor Compression
Modern foundation models such as large language models (LLMs) and large multi-modal models (LMMs) require a massive amount of computational and memory resources. We propose a new framework to convert such LLMs/LMMs into a reduced-dimension latent structure. Our method extends a local activation-aware tensor decomposition to a global attention-aware joint tensor de-composition. Our framework can significantly improve the model accuracy over the existing model compression methods when reducing the latent dimension to realize computationally/memory-efficient LLMs/LLMs. We show the benefit on several benchmark including multi-modal reasoning tasks.
Open AccessLatentLLM: Attention-Aware Joint Tensor Compression
Modern foundation models such as large language models (LLMs) and large multi-modal models (LMMs) require a massive amount of computational and memory resources. We propose a new framework to convert such LLMs/LMMs into a reduced-dimension latent structure. Our method extends a local activation-aware tensor decomposition to a global attention-aware joint tensor de-composition. Our framework can significantly improve the model accuracy over the existing model compression methods when reducing the latent dimension to realize computationally/memory-efficient LLMs/LLMs. We show the benefit on several benchmark including multi-modal reasoning tasks.
- Open Access
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (19) , 20139-20147
Forget to Flourish: Leveraging Machine-Unlearning on Pretrained Language Models for Privacy Leakage
Fine-tuning large language models on private data for downstream applications poses significant privacy risks in potentially exposing sensitive information. Several popular community platforms now offer convenient distribution of a large variety of pre-trained models, allowing anyone to publish without rigorous verification. This scenario creates a privacy threat, as pre-trained models can be intentionally crafted to compromise the privacy of fine-tuning datasets. In this study, we introduce a novel poisoning technique that uses model-unlearning as an attack tool. This approach manipulates a pre-trained language model to increase the leakage of private data during the fine-tuning process. Our method enhances both membership inference and data extraction attacks while preserving model utility. Experimental results across different models, datasets, and fine-tuning setups demonstrate that our attacks significantly surpass baseline performance. This work serves as a cautionary note for users who download pretrained models from unverified sources, highlighting the potential risks involved.
- 11 April 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI Conference on Artificial Intelligence
- Vol. 39 (19) , 20139-20147
Open AccessForget to Flourish: Leveraging Machine-Unlearning on Pretrained Language Models for Privacy Leakage
Fine-tuning large language models on private data for downstream applications poses significant privacy risks in potentially exposing sensitive information. Several popular community platforms now offer convenient distribution of a large variety of pre-trained models, allowing anyone to publish without rigorous verification. This scenario creates a privacy threat, as pre-trained models can be intentionally crafted to compromise the privacy of fine-tuning datasets. In this study, we introduce a novel poisoning technique that uses model-unlearning as an attack tool. This approach manipulates a pre-trained language model to increase the leakage of private data during the fine-tuning process. Our method enhances both membership inference and data extraction attacks while preserving model utility. Experimental results across different models, datasets, and fine-tuning setups demonstrate that our attacks significantly surpass baseline performance. This work serves as a cautionary note for users who download pretrained models from unverified sources, highlighting the potential risks involved.
- Open Access
Smoothed Embeddings for Robust Language Models
Improving the safety and reliability of large language models (LLMs) is a crucial aspect of realizing trustworthy AI systems. Although alignment methods aim to suppress harmful content generation, LLMs are often still vulnerable to jailbreaking attacks that employ adversarial inputs that subvert alignment and induce harmful outputs. We propose the Randomized Embedding Smoothing and Token Aggregation (RESTA) defense, which adds random noise to the embedding vectors and performs aggregation during the generation of each output token, with the aim of better preserving semantic information. Our experiments demonstrate that our approach achieves superior robustness versus utility tradeoffs compared to the baseline defenses.
Open AccessSmoothed Embeddings for Robust Language Models
Improving the safety and reliability of large language models (LLMs) is a crucial aspect of realizing trustworthy AI systems. Although alignment methods aim to suppress harmful content generation, LLMs are often still vulnerable to jailbreaking attacks that employ adversarial inputs that subvert alignment and induce harmful outputs. We propose the Randomized Embedding Smoothing and Token Aggregation (RESTA) defense, which adds random noise to the embedding vectors and performs aggregation during the generation of each output token, with the aim of better preserving semantic information. Our experiments demonstrate that our approach achieves superior robustness versus utility tradeoffs compared to the baseline defenses.
- Open Access
- 2 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 3466-3480
Analyzing Inference Privacy Risks Through Gradients In Machine Learning
In distributed learning settings, models are iteratively updated with shared gradients computed from potentially sensitive user data. While previous work has studied various privacy risks of sharing gradients, our paper aims to provide a systematic approach to analyze private information leakage from gradients. We present a unified game-based framework that encompasses a broad range of attacks including attribute, property, distributional, and user disclosures. We investigate how different uncertainties of the adversary affect their inferential power via extensive experiments on five datasets across various data modalities. Our results demonstrate the inefficacy of solely relying on data aggregation to achieve privacy against inference attacks in distributed learning. We further evaluate five types of defenses, namely, gradient pruning, signed gradient descent, adversarial perturbations, variational information bottleneck, and differential privacy, under both static and adaptive adversary settings. We provide an information-theoretic view for analyzing the effectiveness of these defenses against inference from gradients. Finally, we introduce a method for auditing attribute inference privacy, improving the empirical estimation of worst-case privacy through crafting adversarial canary records.
- 2 December 2024
- Published by Association for Computing Machinery (ACM)
- p. 3466-3480
Open AccessAnalyzing Inference Privacy Risks Through Gradients In Machine Learning
In distributed learning settings, models are iteratively updated with shared gradients computed from potentially sensitive user data. While previous work has studied various privacy risks of sharing gradients, our paper aims to provide a systematic approach to analyze private information leakage from gradients. We present a unified game-based framework that encompasses a broad range of attacks including attribute, property, distributional, and user disclosures. We investigate how different uncertainties of the adversary affect their inferential power via extensive experiments on five datasets across various data modalities. Our results demonstrate the inefficacy of solely relying on data aggregation to achieve privacy against inference attacks in distributed learning. We further evaluate five types of defenses, namely, gradient pruning, signed gradient descent, adversarial perturbations, variational information bottleneck, and differential privacy, under both static and adaptive adversary settings. We provide an information-theoretic view for analyzing the effectiveness of these defenses against inference from gradients. Finally, we introduce a method for auditing attribute inference privacy, improving the empirical estimation of worst-case privacy through crafting adversarial canary records.
- 1 December 2024
- Published by IOP Publishing in Journal of Neural Engineering
- Vol. 21 (6) , 066031
Improving subject transfer in EEG classification with divergence estimation
Objective. Classification models for electroencephalogram (EEG) data show a large decrease in performance when evaluated on unseen test subjects. We improve performance using new regularization techniques during model training. Approach. We propose several graphical models to describe an EEG classification task. From each model, we identify statistical relationships that should hold true in an idealized training scenario (with infinite data and a globally-optimal model) but that may not hold in practice. We design regularization penalties to enforce these relationships in two stages. First, we identify suitable proxy quantities (divergences such as Mutual Information and Wasserstein-1) that can be used to measure statistical independence and dependence relationships. Second, we provide algorithms to efficiently estimate these quantities during training using secondary neural network models. Main results. We conduct extensive computational experiments using a large benchmark EEG dataset, comparing our proposed techniques with a baseline method that uses an adversarial classifier. We first show the performance of each method across a wide range of hyperparameters, demonstrating that each method can be easily tuned to yield significant benefits over an unregularized model. We show that, using ideal hyperparameters for all methods, our first technique gives significantly better performance than the baseline regularization technique. We also show that, across hyperparameters, our second technique gives significantly more stable performance than the baseline. The proposed methods require only a small computational cost at training time that is equivalent to the cost of the baseline. Significance. The high variability in signal distribution between subjects means that typical approaches to EEG signal modeling often require time-intensive calibration for each user, and even re-calibration before every use. By improving the performance of population models in the most stringent case of zero-shot subject transfer, we may help reduce or eliminate the need for model calibration.
- 1 December 2024
- Published by IOP Publishing in Journal of Neural Engineering
- Vol. 21 (6) , 066031
Improving subject transfer in EEG classification with divergence estimation
Objective. Classification models for electroencephalogram (EEG) data show a large decrease in performance when evaluated on unseen test subjects. We improve performance using new regularization techniques during model training. Approach. We propose several graphical models to describe an EEG classification task. From each model, we identify statistical relationships that should hold true in an idealized training scenario (with infinite data and a globally-optimal model) but that may not hold in practice. We design regularization penalties to enforce these relationships in two stages. First, we identify suitable proxy quantities (divergences such as Mutual Information and Wasserstein-1) that can be used to measure statistical independence and dependence relationships. Second, we provide algorithms to efficiently estimate these quantities during training using secondary neural network models. Main results. We conduct extensive computational experiments using a large benchmark EEG dataset, comparing our proposed techniques with a baseline method that uses an adversarial classifier. We first show the performance of each method across a wide range of hyperparameters, demonstrating that each method can be easily tuned to yield significant benefits over an unregularized model. We show that, using ideal hyperparameters for all methods, our first technique gives significantly better performance than the baseline regularization technique. We also show that, across hyperparameters, our second technique gives significantly more stable performance than the baseline. The proposed methods require only a small computational cost at training time that is equivalent to the cost of the baseline. Significance. The high variability in signal distribution between subjects means that typical approaches to EEG signal modeling often require time-intensive calibration for each user, and even re-calibration before every use. By improving the performance of population models in the most stringent case of zero-shot subject transfer, we may help reduce or eliminate the need for model calibration.