“Working alongside Clodéric during two years has been a pleasure. Team player, he is an excellent mentor and colleague.”
About
Experience & Education
Licenses & Certifications
Publications
-
Hiking up that HILL with Cogment-Verse: Train & Operate Multi-agent Systems Learning from Humans
AAMAS 2023
As more AI systems are deployed, humans are increasingly required to interact with them in multiple settings. However, such AI sys- tems seldom learn from these interactions with humans, which provides an important opportunity to improve from human ex- pertise and context awareness. Several recent results in the fields of reinforcement learning (RL) and human-in-the-loop learning (HILL) show that AI agents can perform better when humans are involved in their training process. Humans can provide…
As more AI systems are deployed, humans are increasingly required to interact with them in multiple settings. However, such AI sys- tems seldom learn from these interactions with humans, which provides an important opportunity to improve from human ex- pertise and context awareness. Several recent results in the fields of reinforcement learning (RL) and human-in-the-loop learning (HILL) show that AI agents can perform better when humans are involved in their training process. Humans can provide rewards to the agent, demonstrate tasks, design curricula, or act directly in the environment, but these potential performance improvements also come with architectural, functional design, and engineering com- plexities. This paper discusses Cogment, a unifying open-source framework that introduces a formalism to support a variety of human(s)-agent(s) collaboration topologies and training approaches. Cogment addresses the complexity of training with humans within a production-ready platform. On top of Cogment, we introduce Cogment Verse a research platform dedicated to the research com- munity to facilitate the implementation of HILL and Multi-Agent RL experiments. With these platforms, our end goal is to enable the generalization of intelligence ecosystems where AI agents and humans learn from each other and collaborate to address increasingly complex or sensitive use cases. The video demonstration is available at https://youtu.be/v-K0DqIL9K0
Other authorsSee publication -
WIP: Human-AI interactions in real-world complex environments using a comprehensive reinforcement learning framework
AAMAS 2023 ALA Workshop
Deep reinforcement learning (RL) has successfully tackled many real-world tasks. However, these algorithms suffer from the well-known sample-inefficiency problem. Deep RL systems usually require millions of environment interactions to learn and have stable performance. In this work, we show that human-AI teams outperform human-only controlled and fully autonomous teams for complex tasks. We develop a novel simulator for a critical infrastruc- ture scenario and a user interface for humans to…
Deep reinforcement learning (RL) has successfully tackled many real-world tasks. However, these algorithms suffer from the well-known sample-inefficiency problem. Deep RL systems usually require millions of environment interactions to learn and have stable performance. In this work, we show that human-AI teams outperform human-only controlled and fully autonomous teams for complex tasks. We develop a novel simulator for a critical infrastruc- ture scenario and a user interface for humans to effectively advise AI agents. We show that humans can provide useful advice to the RL agents, allowing them to improve learning in a multi-agent setting.
Other authorsSee publication -
Cogment: Open Source Framework For Distributed Multi-actor Training, Deployment & Operations
arXiv
Involving humans directly for the benefit of AI agents' training is getting traction thanks to several advances in reinforcement learning and human-in-the-loop learning. Humans can provide rewards to the agent, demonstrate tasks, design a curriculum, or act in the environment, but these benefits also come with architectural, functional design and engineering complexities. We present Cogment, a unifying open-source framework that introduces an actor formalism to support a variety of…
Involving humans directly for the benefit of AI agents' training is getting traction thanks to several advances in reinforcement learning and human-in-the-loop learning. Humans can provide rewards to the agent, demonstrate tasks, design a curriculum, or act in the environment, but these benefits also come with architectural, functional design and engineering complexities. We present Cogment, a unifying open-source framework that introduces an actor formalism to support a variety of humans-agents collaboration typologies and training approaches. It is also scalable out of the box thanks to a distributed micro service architecture, and offers solutions to the aforementioned complexities.
Other authorsSee publication -
The three stages of Explainable AI: How explainability facilitates real world deployment of AI
Humains et IA (HIA) workshop at the 20th Extraction et Gestion des Connaissances (EGC)
Explainable AI has recently seen a renewed interest. We believe these techniques make a true difference when it comes to deploying AIs, especially in the entreprise world. In this article we introduce a framework categorizing explainability levels, their impact on operationalized AI and their requirements.
Other authorsSee publication -
Periodic split method: learning more readable decision trees for human activities
Conférence Nationale sur les Applications Pratiques de l’Intelligence Artificielle
Placing your trust in algorithms is a major issue in society today. This article introduces a novel split method for decision tree generation algorithms aimed at improving the quality/readability ratio of generated decision trees. We focus on human activities learning that allow the definition of new temporal features. By virtue of these features, we present here the periodic split method, which produces similar or superior quality trees with reduced tree depth.
Other authorsSee publication -
Environmentally Conscious AI: Improving Spatial Analysis and Reasoning
GDC '14, AI Summit
As game worlds become more complex, game characters need to become more "aware" of their surroundings. These days, it is no longer sufficiently believable for an agent to merely walk through these environments. Believable movement demands so much more. Characters may now need to jump, climb, duck or vault in order to navigate through their world. Additionally, simply "being" in a space brings in issues with cover selection, visibility, positioning and other forms of spatial awareness. In the…
As game worlds become more complex, game characters need to become more "aware" of their surroundings. These days, it is no longer sufficiently believable for an agent to merely walk through these environments. Believable movement demands so much more. Characters may now need to jump, climb, duck or vault in order to navigate through their world. Additionally, simply "being" in a space brings in issues with cover selection, visibility, positioning and other forms of spatial awareness. In the past, this often meant time-consuming manual markup of levels or highly constrained level designs, and it certainly didn't support dynamic environments! This session will show two different architectures being developed - one by Havok and one by MASA - to help automate those spatial issues in a way that is independent of level design.
Languages
-
English
Full professional proficiency
-
French
Native or bilingual proficiency
Recommendations received
3 people have recommended Clodéric
Join now to viewOther similar profiles
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More