Nous utilisons des témoins pour analyser le trafic et l’utilisation de notre site web, afin de personnaliser votre expérience. Vous pouvez désactiver ces technologies à tout moment, mais cela peut restreindre certaines fonctionnalités du site. Consultez notre Politique de protection de la vie privée pour en savoir plus.
Paramètre des cookies
Vous pouvez activer et désactiver les types de cookies que vous souhaitez accepter. Cependant certains choix que vous ferez pourraient affecter les services proposés sur nos sites (ex : suggestions, annonces personnalisées, etc.).
Cookies essentiels
Ces cookies sont nécessaires au fonctionnement du site et ne peuvent être désactivés. (Toujours actif)
Cookies analyse
Acceptez-vous l'utilisation de cookies pour mesurer l'audience de nos sites ?
Multimedia Player
Acceptez-vous l'utilisation de cookies pour afficher et vous permettre de regarder les contenus vidéo hébergés par nos partenaires (YouTube, etc.) ?
Publications
Multi-modal Decoding of Reach-to-Grasping from EEG and EMG via Neural Networks
Relative biological effectiveness of clinically relevant photon energies for the survival of human colorectal, cervical, and prostate cancer cell lines
Self-correction is a highly desirable capability of large language models (LLMs), yet it has consistently been found to be largely ineffecti… (voir plus)ve in modern LLMs. Current methods for training self-correction typically depend on either multiple models, a more advanced model, or additional forms of supervision. To address these shortcomings, we develop a multi-turn online reinforcement learning (RL) approach, SCoRe, that significantly improves an LLM's self-correction ability using entirely self-generated data. To build SCoRe, we first show that variants of supervised fine-tuning (SFT) on offline model-generated correction traces are often insufficient for instilling self-correction behavior. In particular, we observe that training via SFT falls prey to either a distribution mismatch between mistakes made by the data-collection policy and the model's own responses, or to behavior collapse, where learning implicitly prefers only a certain mode of correction behavior that is often not effective at self-correction on test problems. SCoRe addresses these challenges by training under the model's own distribution of self-generated correction traces and using appropriate regularization to steer the learning process into learning a self-correction behavior that is effective at test time as opposed to fitting high-reward responses for a given prompt. This regularization process includes an initial phase of multi-turn RL on a base model to generate a policy initialization that is less susceptible to collapse, followed by using a reward bonus to amplify self-correction. With Gemini 1.0 Pro and 1.5 Flash models, we find that SCoRe achieves state-of-the-art self-correction performance, improving the base models' self-correction by 15.6% and 9.1% respectively on MATH and HumanEval.
Self-correction is a highly desirable capability of large language models (LLMs), yet it has consistently been found to be largely ineffecti… (voir plus)ve in modern LLMs. Current methods for training self-correction typically depend on either multiple models, a more advanced model, or additional forms of supervision. To address these shortcomings, we develop a multi-turn online reinforcement learning (RL) approach, SCoRe, that significantly improves an LLM's self-correction ability using entirely self-generated data. To build SCoRe, we first show that variants of supervised fine-tuning (SFT) on offline model-generated correction traces are often insufficient for instilling self-correction behavior. In particular, we observe that training via SFT falls prey to either a distribution mismatch between mistakes made by the data-collection policy and the model's own responses, or to behavior collapse, where learning implicitly prefers only a certain mode of correction behavior that is often not effective at self-correction on test problems. SCoRe addresses these challenges by training under the model's own distribution of self-generated correction traces and using appropriate regularization to steer the learning process into learning a self-correction behavior that is effective at test time as opposed to fitting high-reward responses for a given prompt. This regularization process includes an initial phase of multi-turn RL on a base model to generate a policy initialization that is less susceptible to collapse, followed by using a reward bonus to amplify self-correction. With Gemini 1.0 Pro and 1.5 Flash models, we find that SCoRe achieves state-of-the-art self-correction performance, improving the base models' self-correction by 15.6% and 9.1% respectively on MATH and HumanEval.
Representing Positional Information in Generative World Models for Object Manipulation
Stefano Ferraro
Pietro Mazzaglia
Tim Verbelen
Bart Dhoedt
Sai Rajeswar
Object manipulation capabilities are essential skills that set apart embodied agents engaging with the world, especially in the realm of rob… (voir plus)otics. The ability to predict outcomes of interactions with objects is paramount in this setting. While model-based control methods have started to be employed for tackling manipulation tasks, they have faced challenges in accurately manipulating objects. As we analyze the causes of this limitation, we identify the cause of underperformance in the way current world models represent crucial positional information, especially about the target's goal specification for object positioning tasks. We introduce a general approach that empowers world model-based agents to effectively solve object-positioning tasks. We propose two declinations of this approach for generative world models: position-conditioned (PCP) and latent-conditioned (LCP) policy learning. In particular, LCP employs object-centric latent representations that explicitly capture object positional information for goal specification. This naturally leads to the emergence of multimodal capabilities, enabling the specification of goals through spatial coordinates or a visual goal. Our methods are rigorously evaluated across several manipulation environments, showing favorable performance compared to current model-based control approaches.
Web applications, accessible via web browsers over the Internet, facilitate complex functionalities without local software installation. In … (voir plus)the context of web applications, a workload refers to the number of user requests sent by users or applications to the underlying system. Existing studies have leveraged web application workloads to achieve various objectives, such as workload prediction and auto-scaling. However, these studies are conducted in an ad hoc manner, lacking a systematic understanding of the characteristics of web application workloads. In this study, we first conduct a systematic literature review to identify and analyze existing studies leveraging web application workloads. Our analysis sheds light on their workload utilization, analysis techniques, and high-level objectives. We further systematically analyze the characteristics of the web application workloads identified in the literature review. Our analysis centers on characterizing these workloads at two distinct temporal granularities: daily and weekly. We successfully identify and categorize three daily and three weekly patterns within the workloads. By providing a statistical characterization of these workload patterns, our study highlights the uniqueness of each pattern, paving the way for the development of realistic workload generation and resource provisioning techniques that can benefit a range of applications and research areas.
We study the problem of building reasoning agents that are able to generalize in an effective manner. Towards this goal, we propose an end-t… (voir plus)o-end approach for building model-based reinforcement learning agents that dynamically focus their reasoning to the relevant aspects of the environment: after automatically identifying the distinct aspects of the environment, these agents dynamically filter out the relevant ones and then pass them to their simulator to perform partial reasoning. Unlike existing approaches, our approach works with pixel-based inputs and it allows for interpreting the focal points of the agent. Our quantitative analyses show that the proposed approach allows for effective generalization in high-dimensional domains with raw observational inputs. We also perform ablation analyses to validate of design choices. Finally, we demonstrate through qualitative analyses that our approach actually allows for building agents that focus their reasoning on the relevant aspects of the environment.
We study the problem of building reasoning agents that are able to generalize in an effective manner. Towards this goal, we propose an end-t… (voir plus)o-end approach for building model-based reinforcement learning agents that dynamically focus their reasoning to the relevant aspects of the environment: after automatically identifying the distinct aspects of the environment, these agents dynamically filter out the relevant ones and then pass them to their simulator to perform partial reasoning. Unlike existing approaches, our approach works with pixel-based inputs and it allows for interpreting the focal points of the agent. Our quantitative analyses show that the proposed approach allows for effective generalization in high-dimensional domains with raw observational inputs. We also perform ablation analyses to validate our design choices. Finally, we demonstrate through qualitative analyses that our approach actually allows for building agents that focus their reasoning on the relevant aspects of the environment.
Ultrasound Localization Microscopy (ULM) is a novel super-resolution imaging technique that can image the vasculature in vivo at depth with … (voir plus)resolution far beyond the conventional limit of diffraction. By relying on the localization and tracking of clinically approved microbubbles injected in the blood stream, ULM can provide not only anatomical visualization but also hemodynamic quantification of the microvasculature of different tissues. Various deep-learning approaches have been proposed to address challenges in ULM including denoising, improving microbubble localization, estimating blood flow velocity or performing aberration correction. Proposed deep learning methods often outperform their conventional counterparts by improving image quality and reducing processing time. In addition, their robustness to high concentrations of microbubbles can lead to reduced acquisition times in ULM, addressing a major hindrance to ULM clinical application. Herein, we propose a comprehensive review of the diversity of deep learning applications in ULM focusing on approaches assuming a sparse microbubbles distribution. We first provide an overview of how existing studies vary in the constitution of their datasets or in the tasks targeted by deep learning model. We also take a deeper look into the numerous approaches that have been proposed to improve the localization of microbubbles since they differ highly in their formulation of the optimization problem, their evaluation, or their network architectures. We finally discuss the current limitations and challenges of these methods, as well as the promises and potential of deep learning for ULM in the future.
2024-09-17
IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control (publié)
Ultrasound localization microscopy (ULM) is a novel super-resolution imaging technique that can image the vasculature in vivo at depth with … (voir plus)resolution far beyond the conventional limit of diffraction. By relying on the localization and tracking of clinically approved microbubbles injected in the blood stream, ULM can provide not only anatomical visualization but also hemodynamic quantification of the microvasculature. Several deep learning approaches have been proposed to address challenges in ULM including denoising, improving microbubble localization, estimating blood flow velocity, or performing aberration correction. Proposed deep learning methods often outperform their conventional counterparts by improving image quality and reducing processing time. In addition, their robustness to high concentrations of microbubbles can lead to reduced acquisition times in ULM, addressing a major hindrance to ULM clinical application. Herein, we propose a comprehensive review of the diversity of deep learning applications in ULM focusing on approaches assuming a sparse microbubble distribution. We first provide an overview of how existing studies vary in the constitution of their datasets or in the tasks targeted by the deep learning model. We also take a deeper look into the numerous approaches that have been proposed to improve the localization of microbubbles since they differ highly in their formulation of the optimization problem, their evaluation, or their network architectures. We finally discuss the current limitations and challenges of these methods, as well as the promises and potential of deep learning for ULM in the future.
2024-09-17
IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control (publié)
Teacher-Student Curriculum Learning (TSCL) is a curriculum learning framework that draws inspiration from human cultural transmission and le… (voir plus)arning. It involves a teacher algorithm shaping the learning process of a learner algorithm by exposing it to controlled experiences. Despite its success, understanding the conditions under which TSCL is effective remains challenging. In this paper, we propose a data-centric perspective to analyze the underlying mechanics of the teacher-student interactions in TSCL. We leverage cooperative game theory to describe how the composition of the set of experiences presented by the teacher to the learner, as well as their order, influences the performance of the curriculum that is found by TSCL approaches. To do so, we demonstrate that for every TSCL problem, there exists an equivalent cooperative game, and several key components of the TSCL framework can be reinterpreted using game-theoretic principles. Through experiments covering supervised learning, reinforcement learning, and classical games, we estimate the cooperative values of experiences and use value-proportional curriculum mechanisms to construct curricula, even in cases where TSCL struggles. The framework and experimental setup we present in this work represent a novel foundation for a deeper exploration of TSCL, shedding light on its underlying mechanisms and providing insights into its broader applicability in machine learning.
