A5053929522- Natural Language Processing Techniques
- Topic Modeling
- Authorship Attribution and Profiling
- Speech and dialogue systems
- Spam and Phishing Detection
- Text and Document Classification Technologies
- EEG and Brain-Computer Interfaces
- Speech Recognition and Synthesis
- Semantic Web and Ontologies
- Advanced Text Analysis Techniques
- Web Data Mining and Analysis
- Hate Speech and Cyberbullying Detection
- Blind Source Separation Techniques
- Advanced Image and Video Retrieval Techniques
- Sentiment Analysis and Opinion Mining
- Spanish Linguistics and Language Studies
- Music and Audio Processing
- Data Management and Algorithms
- Misinformation and Its Impacts
- Image Retrieval and Classification Techniques
- Mental Health via Writing
- Emotion and Mood Recognition
- Speech and Audio Processing
- Cultural and political discourse analysis
- Phonetics and Phonology Research
National Institute of Astrophysics, Optics and Electronics
2016-2025
Université d'Artois
2019-2020
Consejo Nacional de Ciencia y Tecnología
2008-2018
Twitter (United States)
2018
Google (United States)
2018
Inria Rennes - Bretagne Atlantique Research Centre
2018
Centro Médico ABC
2017
Artistic Realization Technologies
2017
Universidad Michoacana de San Nicolás de Hidalgo
2011
Universidad de Puebla
2011
We present Theseus, an efficient application-agnostic open source library for differentiable nonlinear least squares (DNLS) optimization built on PyTorch, providing a common framework end-to-end structured learning in robotics and vision. Existing DNLS implementations are application specific do not always incorporate many ingredients important efficiency. Theseus is application-agnostic, as we illustrate with several example applications that using the same underlying components, such...
Adversarial attacks deliberately modify deep learning inputs, mislead models, and cause incorrect results. Previous adversarial on sentiment analysis models have demonstrated success in misleading these models. However, most existing applied a generalized approach to input modifications, without considering the characteristics objectives of different levels. Specifically, for aspect-based analysis, there is lack attack methods that inputs accordance with evaluated aspects. Consequently,...
Pain is considered an unpleasant but vital experience for every living being, it extremely complex and subjective because composed of different variables related to their experiences. Their history, biological sex, socio/cultural context, mood, hormonal changes can affect perception. Nociceptive pain more linked tissue damage or stimulus, this will always have a reaction that goes from its activation in the nociceptors peripheral nerves being central nervous system. The most common way...
Depression is a disease that affects considerable portion of the world population. Severe cases depression interfere with common live patients, for those patients strict monitoring necessary in order to control progress and prevent undesired side effects. A way keep track by means online via human-computer-interaction. The AVEC'14 challenge aims at developing technology towards patients. This paper describes an approach recognition from audiovisual information context challenge. proposed...
Large-scale, data-intensive scientific applications are often expressed as workflows (SWfs). In this paper, we consider the problem of efficient scheduling a large SWf in multisite cloud, i.e., cloud with geo-distributed data centers (sites). The reasons for using multiple sites to run that is already distributed, necessary resources exceed limits at single site, or monetary cost lower. metadata management has critical impact on efficiency it provides global view location and enables task...
This paper examines the importance of different groups speech acoustic features in estimation emotional primitives which define a three-dimensional continuous model emotions. A set proposed is extracted from database German spontaneous speech. tries to represent several aspects content that have been discussed separately other works. Features selection and dimensionality reduction techniques are applied find subsets best estimate Valence, Activation Dominance. Finally, estimated using...
This work aims to interpret the EEG signals associated with actions imagine pronunciation of words that belong a reduced vocabulary without moving articulatory muscles and uttering any audible sound (unspoken speech). Specifically, reflects movements control cursor on computer. We have recorded from 21 subjects using markers based basic protocol. The discrete wavelet transform (DWT) is used extract features delimited windows, subset them frequency ranges below 32 Hz further selected. These...