A5056033969- Topic Modeling
- Natural Language Processing Techniques
- Advanced Image and Video Retrieval Techniques
- Multimodal Machine Learning Applications
- 3D Surveying and Cultural Heritage
- Advanced Text Analysis Techniques
- Indoor and Outdoor Localization Technologies
- Sentiment Analysis and Opinion Mining
- Domain Adaptation and Few-Shot Learning
- Video Analysis and Summarization
- 3D Shape Modeling and Analysis
- Human Mobility and Location-Based Analysis
- Speech and Audio Processing
- Manufacturing Process and Optimization
- Text and Document Classification Technologies
Université de Montréal
2023-2024
Mila - Quebec Artificial Intelligence Institute
2023
University of Alberta
2019
Conversational search supports multi-turn user-system interactions to solve complex information needs. Compared with the traditional single-turn ad-hoc search, conversational faces a more intent understanding problem because session is much longer and contains many noisy tokens. However, existing dense retrieval solutions simply fine-tune pre-trained query encoder on limited data, which are hard achieve satisfactory performance in such scenario. Meanwhile, learned latent representation also...
Conversational query rewriting (CQR) realizes conversational search by reformulating the dialogue into a standalone rewrite. However, existing CQR models either are not learned toward improving downstream performance or inefficiently generate rewrite token-by-token from scratch while neglecting fact that often has large overlap with In this paper, we propose EdiRCS, new text editing-based model tailored for search. most of tokens selected in non-autoregressive fashion and only few generated...
Opinion mining aims to detect and extract relevant information from large quantity of customer reviews. Automatic opinion summarization then seeks create a consensual point view often oriented toward the main sentiment clients render their experience. Although factual is valuable for companies understand what works or not in products, approaches that convey objectivity, constructive feedback reviews have yet be explored. We propose an adversarial multi-task learning model document address...
Current text-video retrieval methods mainly rely on cross-modal matching between queries and videos to calculate their similarity scores, which are then sorted obtain results. This method considers the each candidate video query, but it incurs a significant time cost will increase notably with of candidates. Generative models common in natural language processing computer vision, have been successfully applied document retrieval, application multimodal remains unexplored. To enhance...
Reconstructing fine-grained spatial densities from coarse-grained measurements, namely the aggregate observations recorded for each subregion in field of interest, is a critical problem many real world applications. In this paper, we propose novel Constrained Spatial Smoothing (CSS) approach data reconstruction. We observe that local continuity exists types data. Based on observation, our performs sparse recovery via finite element method, while meantime enforcing aggregated observation...
Question understanding is an important issue to the success of a Knowledge-based Answering (KBQA) system.However, existing study does not pay enough attention this given that questions in KBQA datasets are usually expressed simple and straightforward way. This line with actual linguistic conventions, which often use lot modifiers. To facilitate on evaluating enhancing question ability systems, paper proposes construct complex-modified question-answering (XMQAs) dataset based datasets. With...