Research Interests +
|
|
|
|
Short Profile
 |
Associate Professor, Flowers Team, U2IS laboratory, Ensta-Paris |
 |
Former Associate Professor, Informatics Department, IMT Atlantique |
 |
Former member of the IHSEV team of the CNRS laboratory Lab-STICC |
 |
PhD in Computer Science with highest honors : active learning based on social guidance and artificial curiosity. |
 |
MSc in Computer Science, Ecole Polytechnique, France. |
 |
MEng in Adaptative Machine Systems, Osaka University, Japan. |
Selected Publications +
Reconciling Spatial and Temporal Abstractions for Goal Representation
Mehdi Zadem, Sergio Mover, and Sao Mai Nguyen (2024)
International Conference on Learning Representations, {ICLR}.
[ bib ]
Emergence of a symbolic representation of a continuous state space for hierachical learning by reachability analysis.
Mehdi Zadem, Sergio Mover, and Sao Mai Nguyen (2024)
International Conference on Learning Representations, {ICLR}.
[ bib ]
Emergence of a symbolic representation of a continuous state space for hierachical learning by reachability analysis.
A Medical Low-Back Pain Physical Rehabilitation Dataset for Human Body Movement Analysis
Sao Mai Nguyen, Maxime Devanne, Olivier Remy-Neris, Mathieu Lempereur, and Andre Thepaut (2024)
International Joint Conference on Neural Networks.
[ bib video youtube code ]
First medical dataset of low-back pain patients during week-long clinical trial of physical rehabilitation from the Keraal Project
Sao Mai Nguyen, Maxime Devanne, Olivier Remy-Neris, Mathieu Lempereur, and Andre Thepaut (2024)
International Joint Conference on Neural Networks.
[ bib video youtube code ]
First medical dataset of low-back pain patients during week-long clinical trial of physical rehabilitation from the Keraal Project
Intrinsically Motivated Open-Ended Multi-Task Learning Using Transfer Learning to Discover Task Hierarchy
Nicolas Duminy, Sao Mai Nguyen, Junshuai Zhu, Dominique Duhaut, and Jerome Kerdreux (2021)
Applied Sciences, 11, (3).
[ bib youtube code ]
Active learning combines curiosity with goal-based RL and imitation for hierarchical tasks by automatic curriculum learning.
Nicolas Duminy, Sao Mai Nguyen, Junshuai Zhu, Dominique Duhaut, and Jerome Kerdreux (2021)
Applied Sciences, 11, (3).
[ bib youtube code ]
Active learning combines curiosity with goal-based RL and imitation for hierarchical tasks by automatic curriculum learning.
Generative Pretrained Embedding and Hierarchical Irregular Time Series Representation for Daily Living Activity Recognition
Damien Bouchabou and Sao Mai Nguyen (2024)
ECAI 2024. 4764 - 4771
[ bib code ]
GPT algorithm can learn directly from IoT Sensor signals which are multimodal multivariate irregularly sampled time series, to encode context-aware semantic by self-supervised training.
Damien Bouchabou and Sao Mai Nguyen (2024)
ECAI 2024. 4764 - 4771
[ bib code ]
GPT algorithm can learn directly from IoT Sensor signals which are multimodal multivariate irregularly sampled time series, to encode context-aware semantic by self-supervised training.
Using Language Model to Bootstrap Human Activity Recognition Ambient Sensors Based in Smart Homes
Damien Bouchabou, Sao Mai Nguyen, Christophe Lohr, Benoit LeDuc, and Ioannis Kanellos (2021)
Electronics, 10, (20).
[ bib ]
NLP embeddings with bi-LSTM encode context-aware semantic for long-term dependencies in human activity recognition.
Damien Bouchabou, Sao Mai Nguyen, Christophe Lohr, Benoit LeDuc, and Ioannis Kanellos (2021)
Electronics, 10, (20).
[ bib ]
NLP embeddings with bi-LSTM encode context-aware semantic for long-term dependencies in human activity recognition.
Active choice of teachers, learning strategies and goals for a socially guided intrinsic motivation learner
Sao Mai Nguyen and Pierre-Yves Oudeyer (2012)
Paladyn Journal of Behavioural Robotics, 3, (3). 136-146
[ bib code ]
Robot learner combines goal-oriented RL and imitation learning to actively choose what, when to learn, what, when and whom to imitate for multi-task learning by curiosity-based autonomous curriculum learning.
Sao Mai Nguyen and Pierre-Yves Oudeyer (2012)
Paladyn Journal of Behavioural Robotics, 3, (3). 136-146
[ bib code ]
Robot learner combines goal-oriented RL and imitation learning to actively choose what, when to learn, what, when and whom to imitate for multi-task learning by curiosity-based autonomous curriculum learning.
Socially Guided Intrinsic Motivation for Robot Learning of Motor Skills
Sao Mai Nguyen and Pierre-Yves Oudeyer (2014)
Autonomous Robots, 36, (3). 273-294
[ bib ]
Robot learner combines goal-oriented RL and imitation learning to actively choose what, when to learn, what, when and whom to imitate for parametrized tasks by curiosity-based autonomous curriculum learning.
Sao Mai Nguyen and Pierre-Yves Oudeyer (2014)
Autonomous Robots, 36, (3). 273-294
[ bib ]
Robot learner combines goal-oriented RL and imitation learning to actively choose what, when to learn, what, when and whom to imitate for parametrized tasks by curiosity-based autonomous curriculum learning.
Profile Description
I specialize in cognitive developmental learning, reinforcement learning, imitation learning, automatic curriculum learning for robots : I develop algorithms for robots to learn multi-task controls by designing themselves their curriculum and by active imitation learning. I received my PhD from Inria in 2013 in computer science, for machine learning algorithms combining reinforcement learning and active imitation learning for interactive and multi-task learning. I hold an Engineer degree from Ecole Polytechnique, France and a master's degree in adaptive machine systems from Osaka University, Japan. I have coordinated project KERAAL to enable a robot to coach physical rehabilitation. I have participated in project AMUSAAL, for analysing human activities of daily living, and CPER VITAAL for developing assistive technologies for the elderly and disabled. I am currently an assistant professor at Ensta IP Paris, France and was previously with IMT Atlantique. I also act as an associate editor of the journal IEEE TCDS and the co-chair of the Task force "Action and Perception" of the IEEE Technical Committee on Cognitive and Developmental Systems.
Interested in robotics, artificial intelligence and cognitive science, I try to bridge these 3 aspects in my work, and started working in cognitive developmental robotics. I have special interest in linking research of machine learning, robotic embodiment, child development psychology and neuroscience. I also explore possibilities to make robots adapt to their physical and social environments.
This is why I strive to answer the questions: Can a robot learn like a child? How do children learn? How do human beings or other species learn? Can a robot learn through social interaction with humans?
Old updates up to 2019
- 2019/05/09: I am invited to give a plenary talk and participate in a panel discussion for the seminar on A.I. and IoT : New Horizon, organised by ESATIC under the patronage of the minister of Digital Economy and the Post of Ivory Coast, Claude Isaac De.
- 2019/03/20: Our Team IHSEV is on TV on the French chanel Tebeo
- I am now associate editor of the journal IEEE TCDS
- I am organising with for the journal IEEE TCDS a special issue on Continual Unsupervised Sensorimotor Learning
- Our latest article Duminy et al, 2019, Frontiers in Neurorobotics+ has just been published
- 2018/12/18: Nicolas Duminy received his PhD
- I am member of the program committee of the conference IEEE IRC
- 2018/12/04: I will be presenting at the event EURÊKA, la journée inspirante du Village by CA - "Entreprendre en s'inspirant de la Nature"
- 2018/11/28: I will be presenting at The panel discussion on intelligent robots at the event 360 possibles
- 2018/09/17: I am co-organiser of the Workshop on Continual Unsupervised Sensorimotor Learning at IEEE ICDL-Epirob 2018
- 2018/09/18: Maxime Devanne will present at the EECV workshop Task-cv (Transferring and Adapting Source Knowledge in Computer Vision and VisDA Challenge) Devanne et al, 2018, EECV 2018 Workshops +
- 2018/10/07: Nicolas Duminy will present at IEEE SMC the poster Duminy et al, 2018, IEEE SMC+
- 2017/11/17: Maxime Devanne presented Devanne and Nguyen, 2017, Humanoids on Friday 17th November at Humanoids 2017. +
- 2017/11/04: Maxime Devanne presented our project KERAAL to Sophie CLUZEL, Secretary of State in charge of People with Disabilities, and Julien DENORMANDIE, secretary of state attached to the Minister of Territorial Cohesion, during the inauguration of the research chair Maintien@Domicile
- 2017/02/06: Telecom Bretagne has changed names to "Institut Mines Telecom Atlantique Bretagne-Pays de la Loire"
- 2016/10/01: I am coordinator of EU-funded KERAAL project which proposes a robot coach for physical rehabilitation. See the presentation video
- 2014/10/01: I have moved to Telecom Bretagne, in team IHSEV as an associate professor. I am also researcher at the IHSEV team of the CNRS laboratory LabSTICC The content of this website does not include my research projects there yet.
- 2014/07/28: I added the source code of SGIM for download from the Publications page code.
- 2014/02/02: A new journal article has now been published: Clement Moulin-Frier, Sao Mai Nguyen, and Pierre-Yves Oudeyer (2014), Self-organization of early vocal development in infants and machines: The role of intrinsic motivation. Frontiers in Psychology (Cognitive Science), 4(1006).
![[bibtex]](media/bibtex.gif)
|
|
|
|
|
This site is permanently under construction. New content is higlighted in red.
Click on the + to read more details
Click on the + to read more details
Updates
Twitter link
- 10/05/2025: Proud that Qi will present in 2D human pose estimation #HPE, his pose prior-guided refinement approach based on Neural Distance Fields at @_CVsports @CVPR. The polar coordinate-based representation and metric tackles data scarcity of long-jump videos. 📃https://arxiv.org/abs/2505.03445
- 10/10/2024: On my way to #NeurIPS2024 to present an exploratory extension of our #ICLR2024 paper at @IMOLNeurIPS2024, Compositional Learning & Behavioural ML Workshops. Looking forward to chat about open-ended learning, intrinsic motivation, active imitation, hierarchical learning...
- 30/06/2024: The 1st #MovementAnalysis dataset for #PhysicalRehabilitation with patients data throughout a 3w clinical care with @poppy_project will be presented at #IJCNN #WCCI #WCCI2024 download
- 06/05/2024: Looking forward to discussing about #HierarchicalReinforcementLearning #SymbolEmergence #IntrinsicMotivation and #ContinualLearning with all in front of our poster @iclr_conf Halle B #205, Tue 7 May 4:30 p.m.
- 12/09/2023 : Looking forward to learn about open-ended learning and its links with intrinsic motivation at the Paris IMOL workshop this week https://imolconf2023.github.io where my invited talk justifies why hierarchical actions can be combined with intrinsic motivation to enable open-ended learning
- 10/03/2023 : Can we replace Kinect v2 by a cheaper RGB-camera for human movement analysis ? Our comparison on a medical dataset for #PhysicalRehabilitation points the advantages of Openpose & Blazepose Check Aleksa's poster @HRI_Conference #RobotCoach #RobotPoppy @DL youtube
- 30/10/2022 : @SBiomecanique and @SB_Congres : Merci beaucoup pour votre invitation à présenter nos travaux concernant les études cliniques testant notre robot coach de publication
- 31/03/2022 : Heading to the IMOL Workshop for an invited talk on "Intrinsic motivation to interact with teachers for multi-task learning" on Monday. See you there! #RobotLearning #ml #robotics #HierarchicalLearning #intrinsicmotivation #IA #RL #ImitationLearning publication
- 14/03/2022: So excited. The medical report & patient-therapist evaluation of the #RobotCoach #RobotPoppy for #PhysicalRehabilitation in clinical trial by @remy_neris @CHRU_Brest @MaximeDevanne @MatSimonnet @IMTAtlantique @GenerationRobot is published publication
- 8/3/2022 : To transfer knowledge from easier tasks to complex tasks recursively, can the #robot discover the hierarchy between tasks ? How can human demonstrations be beneficial to an #intrinsicmotivation robot? #RobotLearning #ml #robotics #HierarchicalLearning https://hal.archives-ouvertes.fr/hal-03157219
- 29/11/2021: How #NLP improves Human Activity Recognition in #SmartHome using #IoT sensors? Damien proposes a #DeepLearning algo with #LSTM and ELMo to recognize Activities of Daily Living in @ElectronicsMDPI publication. #ADLs #NeuralNetworks #MachineLearning #AI #TimeSeries
- 17/11/2021: At 📅10:00 on 17/11/2021, I will give a talk on "Active interaction between robots and humans for automatic curriculum learning and assistive robotics" at room 405 of Idiap. You can also attend online via the webcast. Thanks @Idiap_ch for the invitation
- 3/11/2021: Je présenterai notre modélisation de la motivation à interagir comme fonction de récompense de #ReinforcementLearning, pour l'#ImitationLearning et pour un robot coach, 🤖à l'invitation du réseau Robotics research in Nouvelle-Aquitaine 📅29/11/2021 à 13h r4 robotique