Date: Friday, the 13th of October at 9 am, Place: room “salle des thèses” at l’Université d’Avignon, Campus Hannah Arendt (centre-ville). Title: « Hierarchical Imitation and Reinforcement Learning for Multi-Domain Task-Oriented Dialogue Systems ». The defense can be followed through the live link below: https://v-au.univ-avignon.fr/live Abstract: In this Ph.D thesis, we study task-oriented dialogue systems that are systems designed to assist users in completing specific tasks, such as booking a flight or ordering food. They typically rely on reinforcement learning paradigm to model the dialogue that allows the system to reason about the user’s goals and preferences, and to select actions that will lead to the desired outcome. Our focus is specifically on learning from a limited number of interactions that is crucial due to the scarcity and costliness of human interactions. Standard reinforcement learning algorithms typically require a large amount of interaction data to achieve good performance. To address this challenge, we aim to make dialogue systems more sample-efficient in their training. We draw from two main ideas: imitation and hierarchy. Our first contribution explores the integration of imitation with reinforcement learning. We investigate how to effectively use expert demonstrations to extrapolate knowledge with minimal generalisation effort. Our second contribution focuses on Plus d'infos
In the context of team Cornet’s seminars, Willie Kouam (LIA) will present his research work on Asymmetric Centrality Game against Network Epidemic Propagation, on September 29, 2023, at 11:35 in the meeting room.
In the context of team Cornet’s seminars, Wesley Coelho (Pasqal) will present his research work on Solving optimization problems with PASQAL quantum computers, on September 14, 2023, at 11:35 in the meeting room. Abstract: The emergence of quantum devices opens many exciting perspectives in the high-performance computing world. Among other quantum platforms, fully programmable neutral atom devices display unique characteristics and, by better controlling quantum entanglement and superposition, they represent a powerful tool to tackle complex problems and computing challenges. In this talk, Clément de Terrasson and Wesley Coelho will show how PASQAL Quantum Computers are used to address complex optimization problems. They will also propose a workshop where participants will be able to use PASQAL solutions to solve optimization problems.
Date: Friday, June 30, 2023. 2pm Place: thesis room (salle des thèses) at the Hannah Arendt campus. For those unable to attend, here is the BBB link for the video conference: https://v-au.univ-avignon.fr/live/bbb-soutenance-de-these-sarkis-moussa-30-juin-2023/ Title: Architecture and Protocols for Public Safety Users in the 5G Cellular Networks Abstract: Public Safety Networks (PSNs) are wireless communication systems designed to meet the needs of emergency responders, including firefighters, police, and many other Public Safety (PS) agencies. These networks are used to prevent or respond to incidents that pose a threat to people or property. Traditionally, these PSNs were supported by reliable, but low-rate radio technologies that provide limited services such as voice communication among Public Safety Users (PSUs). Consequently, their capability to take advantage of recent developments in wireless networks and broadband applications was restricted. At the forefront of wireless communication technologies, 5th Generation (5G) and beyond Cellular Networks (CNs), are ideal for this purpose due to their advanced infrastructure and tailored techniques developed for broadband services. Their capacity for high data transmission, low latency in data exchange, and ability to support a significant number of connected devices make them perfectly suited to overcome the limitations associated with PSNs. Integrating PSNs into 5G can Plus d'infos
In the context of team Cornet’s seminars, Éric Bourreau (LIRMM) will present his research work on Quantum Computers, a New Information Technology Revolution?, on June 23, 2023, at 11:35 in the meeting room. Abstract: The concept of quantum computing dates back to the late 1980s, and the first quantum algorithms were born in the 1990s. However, the novelty of recent years is the construction of actual quantum machines that are beginning to validate the theory. This seminar will attempt to explain how the definition of QuBits (quantum bits) gives rise to a new paradigm of computation. We will try to understand what computational power is now being offered and at what point quantum optimization could become competitive with ‘classical’ optimization methods.
In the context of team Cornet’s seminars, Paolo Zappala (LIA/Orange) will present his research work on Extensive-form games with perfect information, on June 6, 2023, at 11:35 in the meeting room.
Date: 2nd of May at 2:00 pm. Place: Avignon at the Centre d’Enseignement et de Recherche en Informatique (Ada Lovelace Auditorium) The jury members: Title: Assessment of Speech Intelligibility using Deep Learning: Towards Enhanced Interpretability in Clinical Phonetics. Abstract: Speech intelligibility is an essential component of effective communication. It refers to the degree to which a speaker’s intended message can be understood by a listener. This capacity can be hampered as a consequence of speech disorders, which results in a reduced quality of life for individuals. In the case of Head and Neck Cancer (HNC), speech may be affected due to the presence of tumors in the speech production system, but the main cause of speech impairment is typically the tumor treatment including surgery, radiotherapy, chemotherapy, or a combination of these treatments. In such cases, the evaluation of speech quality is crucial to assess the communication deficit of patients and develop targeted treatment plans. In clinical practice, perceptual measures are considered the gold standard for assessing speech disorders. Although these measures are widely used, they suffer from several limitations, the most important of which is their subjectivity. Consequently, the automatic assessment of speech disorders has emerged as a promising alternative to perceptual Plus d'infos
Date: 26th of April at 2:30pm. Place: Centre d’Enseignement et de Recherche en Informatique (Ada Lovelace auditorium) The jury will consist of: Title: Representing evidence for attribute privacy: Bayesian updating, compositional evidence and calibration. Abstract: Attribute privacy in multimedia technology aims to hide only one or a few personal characteristics, or attributes, of an individual rather than the full identity. To give a few examples, these attributes can be the sex, nationality, or health state of the individual. When the attribute to hide is discrete with a finite number of possible values, the attacker’s belief about the attribute is represented by a discrete probability distribution over the set of possible values. The Bayes’ rule is known as an information acquisition paradigm and tells how the likelihood function is changing the prior belief into a posterior belief. In the binary case—i.e. when there are only two possible values for the attribute—the likelihood function can be written in the form of a Log-Likelihood-Ratio (LLR). This has been known as the weight-of-evidence and is considered a good candidate to inform which hypothesis the data is supporting and how strong. The Bayes’ rule can be written as a sum between the LLR and the log-ratio of Plus d'infos
Nous utilisons des cookies pour vous garantir la meilleure expérience sur notre site web. Si vous continuez à utiliser ce site, nous supposerons que vous en êtes satisfait.
