A5002426346- Semantic Web and Ontologies
- Service-Oriented Architecture and Web Services
- Business Process Modeling and Analysis
- Explainable Artificial Intelligence (XAI)
- Topic Modeling
- Advanced Database Systems and Queries
- Advanced Software Engineering Methodologies
- Time Series Analysis and Forecasting
- Data Management and Algorithms
- Advanced Graph Neural Networks
- Data Quality and Management
- Data Stream Mining Techniques
- Traffic Prediction and Management Techniques
- Machine Learning and Data Classification
- Biomedical Text Mining and Ontologies
- Data Mining Algorithms and Applications
- Domain Adaptation and Few-Shot Learning
- Scientific Computing and Data Management
- Machine Learning in Healthcare
- Adversarial Robustness in Machine Learning
- Natural Language Processing Techniques
- Rough Sets and Fuzzy Logic
- Multimodal Machine Learning Applications
- Recommender Systems and Techniques
- Logic, Reasoning, and Knowledge
Institut national de recherche en informatique et en automatique
2014-2025
JPMorgan Chase & Co (United States)
2023-2024
Thales (Canada)
2017-2022
Thales (France)
2017-2022
Research Centre Inria Sophia Antipolis - Méditerranée
2016-2022
Centre National de la Recherche Scientifique
2022
Centre de Recherche en Informatique
2014-2021
McGill University
2020
Thales (Brazil)
2019
Thales (Portugal)
2019
Understanding black box models has become paramount as systems based on opaque Artificial Intelligence (AI) continue to flourish in diverse real-world applications. In response, Explainable AI (XAI) emerged a field of research with practical and ethical benefits across various domains. This paper highlights the advancements XAI its application scenarios addresses ongoing challenges within XAI, emphasizing need for broader perspectives collaborative efforts. We bring together experts from...
The current hype of Artificial Intelligence (AI) mostly refers to the success machine learning and its sub-domain deep learning. However, AI is also about other areas, such as Knowledge Representation Reasoning, or Distributed AI, i.e., areas that need be combined reach level intelligence initially envisioned in 1950s. Explainable (XAI) now core backup for industry apply products at scale, particularly industries operating with critical systems. This paper reviews XAI not only from a Machine...
Visual Question Answering (VQA) is a challenging task that has received increasing attention from both the computer vision and natural language processing communities. A VQA model combines visual textual features in order to answer questions grounded an image. Current works focus on which are answerable by direct analysis of question image alone. We present concept-aware algorithm, ConceptBert, for require common sense, or basic factual knowledge external structured content. Given language,...
Ranking and optimization of web service compositions represent challenging areas research with significant implications for the realization "Web Services" vision. "Semantic services" use formal semantic descriptions functionality interface to enable automated reasoning over compositions. To judge quality overall composition, example, we can start by calculating similarities between outputs inputs connected constituent services, aggregate these values into a measure composition. This paper...
We predict credit applications with off-the-shelf, interchangeable black-box classifiers and we explain single predictions counterfactual explanations. Counterfactual explanations expose the minimal changes required on input data to obtain a different result e.g., approved vs rejected application. Despite their effectiveness, counterfactuals are mainly designed for changing an undesired outcome of prediction i.e. loan rejected. Counterfactuals, however, can be difficult interpret, especially...
The advent of real-time traffic streaming offers users the opportunity to visualise current conditions and congestion information. However, information highlighting underlying reason for tail-backs remains largely unexplored. Broken lights, an accident, a large concert, or road-works reveal important citizens operators alike. Providing such in requires intelligent mechanisms user interfaces order (i) harness heterogeneous data sources (volume, velocity, variety, veracity) (ii) make derived...
Optimizating semantic Web service compositions is known to be NP-hard, so most approaches restrict the number of services and offer poor scalability. We address scalability issue by selecting which satisfy a set constraints rather than attempting produce an optimal composition. Firstly, we define within innovative extensible quality model designed balance fit (or functional quality) with (QoS) metrics. The criterion evaluates links between description parameters, whilst QoS focuses on...
Recommender systems have been successfully used to address the problem of information overload, where consumers goods and services too many choices overwhelming amount about each choice. Here we focus on service recommendation demonstrate need for using multiple criteria regarding qualities, consider contextual dimensions expected use that service. These two requirements are not considered together by existing recommenders systems, motivating our work an approach which unifies both. To make...
This paper presents STAR-CITY, a system supporting semantic traffic analytics and reasoning for city. which integrates (human machine-based) sensor data using variety of formats, velocities volumes, has been designed to provide insight on historical real-time conditions, all efficient urban planning. Our demonstrates how the severity road congestion can be smoothly analyzed, diagnosed, explored predicted web technologies. We present diagnosis predictive reasoning, both interpreting semantics...
Interpreting the inner function of neural networks is crucial for trustworthy development and deployment these black-box models. Prior interpretability methods focus on correlation-based measures to attribute model decisions individual examples. However, are susceptible noise spurious correlations encoded in during training phase (e.g., biased inputs, overfitting, or misspecification). Moreover, this process has proven result noisy unstable attributions that prevent any transparent...
The semantic web promises to bring automation the areas of service selection, discovery, composition, invocation. In this paper we introduce a means facilitating composition by exploiting matchmaking between parameters (i.e., outputs and inputs) enable their connections interactions. idea is that functions are key components find compatibilities among independently descriptions. To end, our approach extends existing methods (exact, plug-in, subsume, intersection fail) with concept abduction...
The current proliferation of software services means users should be supported when selecting one service out the many which meet their needs. Recommender Systems provide such support for products and conventional services, yet direct application to is not straightforward, because scarcity available user feedback, need fine-tune context intended use. In this article, we address these issues by proposing a semantic content-based recommendation approach that analyzes use effective...
Machine learning explanation can significantly boost machine learning's application in decision making, but the usability of current methods is limited human-centric explanation, especially for transfer learning, an important branch that aims at utilizing knowledge from one domain (i.e., a pair dataset and prediction task) to enhance model training another domain. In this paper, we propose ontology-based approach learning. Three kinds knowledge-based explanatory evidence, with different...