A5063971201- Topic Modeling
- Natural Language Processing Techniques
- Advanced Text Analysis Techniques
- Neural Networks and Applications
- Anomaly Detection Techniques and Applications
- Educational and Psychological Assessments
- Electricity Theft Detection Techniques
- Advanced Graph Neural Networks
- Imbalanced Data Classification Techniques
- Data Quality and Management
- Human Pose and Action Recognition
- Building Energy and Comfort Optimization
- Mineral Processing and Grinding
- Neuroscience, Education and Cognitive Function
- Currency Recognition and Detection
- Digital Imaging for Blood Diseases
Beijing Technology and Business University
2024
University of Washington
2023
The Retrieval-Augmented Generation (RAG) framework enhances Large Language Models (LLMs) by retrieving relevant knowledge to broaden their boundaries and mitigate factual hallucinations stemming from gaps. However, the RAG Framework faces challenges in effective retrieval utilization; invalid or misused will interfere with LLM generation, reducing reasoning efficiency answer quality. Existing methods address these issues decomposing expanding queries, introducing special structures, using...
Action recognition from temporal multivariate sequences of features, such as identifying human actions, is typically approached by supervised training it requires many ground truth annotations to reach high accuracy. Unsupervised methods for the organization into clusters have been introduced, however, continue require associate with actions. The challenges in annotation necessitate an effective classification methodology that minimizes required number labels. Active learning (AL) approaches...
Pre-trained language models are increasingly being used in multi-document summarization tasks. However, these need large-scale corpora for pre-training and domain-dependent. Other non-neural unsupervised approaches mostly rely on key sentence extraction, which can lead to information loss. To address challenges, we propose a lightweight yet effective approach called GLIMMER: Graph LexIcal features based Multi-docuMEnt summaRization approach. It first constructs graph from the source...
Speech brain-computer interfaces aim to decipher what a person is trying say from neural activity alone, restoring communication people with paralysis who have lost the ability speak intelligibly. The Brain-to-Text Benchmark '24 and associated competition was created foster advancement of decoding algorithms that convert text. Here, we summarize lessons learned ending on June 1, 2024 (the top 4 entrants also presented their experiences in recorded webinar). largest improvements accuracy were...