A5024067094- Electric and Hybrid Vehicle Technologies
- Electric Vehicles and Infrastructure
- Advanced Battery Technologies Research
- Genetic Neurodegenerative Diseases
- AI in cancer detection
- Hereditary Neurological Disorders
- Digital Imaging for Blood Diseases
- Cervical Cancer and HPV Research
- COVID-19 diagnosis using AI
- Neurological diseases and metabolism
- Domain Adaptation and Few-Shot Learning
- Vehicle emissions and performance
- Multimodal Machine Learning Applications
- Educational Technology and Pedagogy
Shanghai University of Engineering Science
2024
University of Pennsylvania
2024
Jilin University
2022-2023
The escalating environmental concerns and energy crisis caused by internal combustion engines (ICE) have become unacceptable under regulations the crisis. As a promising alternative solution, multi-power source electric vehicles (MPS-EVs) integrate various clean systems to enhance powertrain efficiency. management strategy (EMS) is plays pivotal role for MPS-EVs maximize efficiency, fuel economy, range. Reinforcement Learning (RL) has emerged as an effective methodology EMS development,...
With the rise of machine learning, reinforcement learning (RL) is gradually applied to energy management strategy (EMS) plug-in hybrid electric vehicle (PHEV). Some old algorithms have also achieved better results by combining with learning. In order learn from advantages previous and explore application potential algorithm, this paper proposes an adaptive hierarchical equivalent consumption minimization (ECMS) knowledge proximal policy optimization (PPO). This system advanced data-driven RL...
The equivalent consumption minimisation strategy (ECMS) solves the problem that dynamic programming (DP) cannot be applied in real-time. Its equivalence factor balances of different power sources and affects maintenance battery state charge (SOC) plug-in hybrid electric vehicles driving cycles. However, a systematic optimisation with known cycles gives optimal choice factor. This paper proposed an intelligent energy management (IEMS), which dynamically adjusts according to demand SOC. IEMS...
Although deep learning-based segmentation models have achieved impressive performance on public benchmarks, generalizing well to unseen environments remains a major challenge. To improve the model's generalization ability new domain during evaluation, test-time training (TTT) is challenging paradigm that adapts source-pretrained model in an online fashion. Early efforts TTT mainly focus image classification task. Directly extending these methods semantic easily experiences unstable adaption...
Histopathology image analysis is the golden standard of clinical diagnosis for Cancers. In doctors daily routine and computer-aided diagnosis, Whole Slide Image (WSI) histopathology tissue used analysis. Because extremely large scale resolution, previous methods generally divide WSI into a number patches, then aggregate all patches within by Multi-Instance Learning (MIL) to make slide-level prediction when developing tools. However, most WSI-MIL models using global-attention without pairwise...
Background: Charcot-Marie-Tooth subtype 2A (CMT2A) is an autosomal dominant single-gene motor sensory neuropathy that caused by mitofusin 2 (Mfn2) mutation. It generally believed CMT2A involves mitochondrial fusion disruption. However, how to ameliorate the loss of membrane in remains unclear. Methods: In vivo and vitro mouse models harboring Mfn2H165R, Mfn2G176S Mfn2R364W mutations were constructed. We first examined damage pathophysiological features Mfn2 mutations. Then, we started with...