A5003791218- Topic Modeling
- Natural Language Processing Techniques
- Advanced Text Analysis Techniques
- Advanced Graph Neural Networks
- Speech Recognition and Synthesis
- Semantic Web and Ontologies
- Open Education and E-Learning
- Machine Learning and Data Classification
- Video Analysis and Summarization
- Recommender Systems and Techniques
- Machine Learning in Healthcare
- Domain Adaptation and Few-Shot Learning
- Online and Blended Learning
- Speech and Audio Processing
- Machine Learning and Algorithms
- Indoor and Outdoor Localization Technologies
- Multimodal Machine Learning Applications
- Graph Theory and Algorithms
- Underwater Vehicles and Communication Systems
Huawei Technologies (United Kingdom)
2024
Orange (France)
2021-2022
Laboratoire d’Informatique et Systèmes
2021-2022
Centre National de la Recherche Scientifique
2021-2022
Aix-Marseille Université
2021-2022
National Central University
2017
Sebastian Gehrmann, Abhik Bhattacharjee, Abinaya Mahendiran, Alex Wang, Alexandros Papangelis, Aman Madaan, Angelina Mcmillan-major, Anna Shvets, Ashish Upadhyay, Bernd Bohnet, Bingsheng Yao, Bryan Wilie, Chandra Bhagavatula, Chaobin You, Craig Thomson, Cristina Garbacea, Dakuo Daniel Deutsch, Deyi Xiong, Di Jin, Dimitra Gkatzia, Dragomir Radev, Elizabeth Clark, Esin Durmus, Faisal Ladhak, Filip Ginter, Genta Indra Winata, Hendrik Strobelt, Hiroaki Hayashi, Jekaterina Novikova, Jenna...
The recent achievements and availability of Large Language Models have paved the road to a new range applications use-cases. Pre-trained language models are now being involved at-scale in many fields where they were until absent from. More specifically, progress made by causal generative has open door using them through textual instructions aka. prompts. Unfortunately, performances these prompts highly dependent on exact phrasing used therefore practitioners need adopt fail-retry strategies....
Knowledge Graph (KG) completion has been excessively studied with a massive number of models proposed for the Link Prediction (LP) task.The main limitation such is their insensitivity to time.Indeed, temporal aspect stored facts often ignored.To this end, more and works consider time as parameter complete KGs.In paper, we first demonstrate that, by simply increasing negative samples, recent ATTH model can achieve competitive or even better performance than state-of-the-art on Temporal KGs...
The massive machine-type communication will be at the core of ambient connectivity, requiring for energy-efficient systems. Earlier studies highlighted efficiency positioning approaches based on channel-state information (CSI) in different environments. Many works limited solution assessment to a single room fully indoor testbed. This article extends application CSI indoor–outdoor detection an unprecedented large area and considers mMTC-oriented long-term evolution fifth-generation Internet...
The task of verbalization RDF triples has known a growth in popularity due to the rising ubiquity Knowledge Bases (KBs). formalism is simple and efficient way store facts at large scale. However, its abstract representation makes it difficult for humans interpret. For this purpose, WebNLG challenge aims promoting automated RDF-to-text generation. We propose leverage pre-trainings from augmented data with Transformer model using augmentation strategy. Our experiment results show minimum...
Although Large Language Models (LLMs) are effective in performing various NLP tasks, they still struggle to handle tasks that require extensive, real-world knowledge, especially when dealing with long-tail facts (facts related entities). This limitation highlights the need supplement LLMs non-parametric knowledge. To address this issue, we analysed effects of different types including textual passage and knowledge graphs (KGs). Since have probably seen majority factual question-answering...
In response to the call for agent-based solutions that leverage ever-increasing capabilities of deep models' ecosystem, we introduce Hive -- a comprehensive solution selecting appropriate models and subsequently planning set atomic actions satisfy end-users' instructions. operates over sets and, upon receiving natural language instructions (i.e. user queries), schedules executes explainable plans actions. These can involve one or more available achieve overall task, while respecting...
Nowadays, E-commerce websites and recommendation systems are so common in our lives. From collected data websites, we found there some repeated products customers' buying history different period. These consuming product will be purchased once again after customers run out of these products. Therefore, system focus on product. The approach used differs from previous by taking into consideration time series purchases. We think is a strong relationship between according to users' behavior...
Evaluation in machine learning is usually informed by past choices, for example which datasets or metrics to use. This standardization enables the comparison on equal footing using leaderboards, but evaluation choices become sub-optimal as better alternatives arise. problem especially pertinent natural language generation requires ever-improving suites of datasets, metrics, and human make definitive claims. To following best model practices easier, we introduce GEMv2. The new version...
Knowledge Graph (KG) completion has been excessively studied with a massive number of models proposed for the Link Prediction (LP) task. The main limitation such is their insensitivity to time. Indeed, temporal aspect stored facts often ignored. To this end, more and works consider time as parameter complete KGs. In paper, we first demonstrate that, by simply increasing negative samples, recent AttH model can achieve competitive or even better performance than state-of-the-art on Temporal...
Text generation from Abstract Meaning Representation (AMR) has substantially benefited the popularized Pretrained Language Models (PLMs). Myriad approaches have linearized input graph as a sequence of tokens to fit PLM tokenization requirements. Nevertheless, this transformation jeopardizes structural integrity and is therefore detrimental its resulting representation. To overcome issue, Ribeiro et al. recently proposed StructAdapt, structure-aware adapter which injects connectivity within...
Sebastien Montella, Alexis Nasr, Johannes Heinecke, Frederic Bechet, Lina M. Rojas Barahona. Proceedings of the 17th Conference European Chapter Association for Computational Linguistics. 2023.
Structured Knowledge has recently emerged as an essential component to support fine-grained Question Answering (QA). In general, QA systems query a Base (KB) detect and extract the raw answers final prediction. However, lacking of context, language generation can offer much informative complete response. this paper, we propose combine power transfer learning advantage entity placeholders produce high-quality verbalization extracted from KB. We claim that such approach is especially...