Brain x Machine Intelligence Lab

Brain x Machine Intelligence Lab @ KAIST

Laboratory for brain and machine intelligence, KAIST

News

[Oct 30, 2024] Our technology has been featured in the KAIST Tech-Fair 2024. For six years in a row!

[June 22, 2024]  Our lab received an ASTRA research grant (news link)! It’s entitled

  • Deep neural experimenter: closed-loop alignment of experimental design, brain model, and decoder for reading out executive control signals in working memory

[Jan 18, 2024]  We have three abstracts accepted to Cosyne 2024. Congratulations!

  • “Simulation-based behavioral profiling by model-guided task optimization and task-guided data generation” - Jae Hoon Shin

  • “Robust and Efficient Grid Code Transformation for Rapid Task Transfer” - Heejun Kim

  • “A Novel Pain Measurement Tool by Modelling Free-operant Foraging Behaviour in Immersive Virtual Reality” - led by Shuangyi Tong (Oxford) and Ben Seymour (Oxford)

[Dec 20, 2023] Year-end parties!

[Dec 20, 2023] HeeJun Kim’s piano recital!

[Oct 10, 2023] Geon Yeong Park’s work has been accepted to NeurIPS 2023. Congratulations!

  • Energy-Based Cross Attention for Bayesian Context Update in Text-to-Image Diffusion Models

[Aug 31, 2023] Our technology has been featured in the KAIST tech fair 2023. For five years in a row!

[June 7, 2023] With generous support from Microsoft Research Asia (MSRA), our lab officially started a collaborative research project on neural value alignment for human-robot cooperation. We are thrilled to work with prominent researchers in MSRA!

[April 24, 2023] Our lab was selected as a SW StarLab (Category: AI) by the Ministry of Science and ICT, Korea. It is a prestigious award recognizing research achievements and potential in the key areas of CS. We will study brain-like System3 reinforcement learning for the next eight years- excited!

(Media coverage: YTN, AI Times, etnews, Choson Biz)

[April 1, 2023]  Our lab moved to the Department of Brain & Cognitive Sciences from the Dept of Bio & Brain Engineering (2015-2023)! Sang Wan Lee will take up a joint appointment shared between the Department of Brain & Cognitive Sciences (primary), the Department of Bio and Brain Engineering, and the Kim Jaechul Graduate School of AI (secondary).

[Mar 30, 2023] #Strawberry party (March), #Lab photo (May)

[Feb 27, 2023]  Our lab alumni, Dr. Yujin Cha, won the Sejong science fellowship. Congratulations! He will also work as a research assistant professor beginning in March 2023. His appointment is shared between the KAIST Graduate School of Medical Science and Engineering (GSMSE) and our research center (CNAI).

[Jan 25, 2023]  Two abstracts have been accepted to Cosyne 2023. Congratulations!

  • ”Uncertainty-robust goal embedding in the prefrontal cortex for flexibly stable learning” - Yoondo Sung

  • ”Controlling human cortical and striatal reinforcement learning with meta prediction error” - Jaehoon Shin

[Oct 28, 2022] Our technologies have been featured in the KAIST core tech fair, including this year’s event.

[June 12, 2022] We presented five posters at RLDM 2022.

  • “Uncertainty and goal embeddings in the lateral prefrontal cortex guide flexible and stable reinforcement learning” - Yundo Sung

  • “How human metacognitive exploration improves reinforcement learning in a sparse reward environment” - Su Jin An

  • “Information Amplification in Human-AI Interactions via Reinforcement Learning” - Yujin Cha

  • “Exploring essential computations underlying generalizable human reinforcement learning” - Dongjae Kim (now a postdoc at NYU)

  • “Memory-guided goal-driven reinforcement learning explains subclinical depression” - Gyubin Lee and Minsu Abel Yang

[Feb 1, 2022]  We have three abstracts accepted to Cosyne 2022. Congratulations!

  • ”Rethinking Tolman's latent learning with metacognitive exploration” - Su Jin An

  • ”In silico manipulation of human cortical computation underlying goal-directed learning” - Jaehoon Shin

  • ”Balancing safety and efficiency in human decision-making” - led by Pranav Mahajan (Oxford) and Ben Seymour (Oxford)

[Dec 23, 2021] BML annual symposium. Thank you so much for your participation!

[Dec 22, 2021] In memory of Sang Wan Lee’s PhD father, (Late) Z. Zenn Bien: Persons of distinguished service to science and technology (대한민국 과학기술유공자) Link

[Nov 1, 2021] Dr. Anil Yaman left our lab to start his new lab as an assistant professor at Vrije Universiteit Amsterdam (Free University of Amsterdam). Congratulation - we wish you all the best of luck!

[Oct, 2021]  Two papers were accepted to ICCV 2021 (1 main, 1 oral). One paper will be presented at NeurIPS 2021 WHMD. Congratulations!

[June 7, 2021]  ​Sang Wan Lee (PI) was selected as the recipient of the 2021 IBM Global University Program Academic Award. The award recognizes individual faculty members whose emerging science and technology contains significant interest for universities and IBM​.​ ​IBM provides a gift of $40,000 to the recipient’s institution in recognition of the selection of the project​​.

[Jan 19, 2021]  We have two abstracts accepted to Cosyne 2021. Congratulations!

  • ”Metacognition guides near-optimal exploration of a large state space with sparse rewards” - Su Jin An

  • ”Midbrain dopamine activity during reinforcement learning reflects bias-variance tradeoff” - Dr. Minryung Song.

[DEC, 2020] Two papers has been accepted to EMNLP 2020 (acceptance rate = 22.4%) and AAAI 2021 (acceptance rate = 21%). Congratulations!

  • “F^2-Softmax: Diversifying Neural Text Generation via Frequency Factorized Softmax,” (EMNLP 2020) - collaboration with Humelo

  • "Human Uncertainty Inference via Deterministic Ensemble Neural Networks" (AAAI 2021)

[Aug 28, 2020] Two outstanding paper awards at 2020 Korean Artificial Intelligent Association conference. Congratulations!

  • Jae Hoon Shin (PhD student) received the outstanding paper award. The paper is entitled “Deep Interaction between Reinforcement Learning Algorithms and Human Reinforcement Learning”.

  • Minsu Abel Yang (undergraduate intern) also received the outstanding paper award. The paper is entitled “Biological Reinforcement Learning via Predictive Spacetime Encoding”.

[Mar 2, 2020] Our work has been featured in the 2020 spring edition of KAIST breakthroughs. (the article “Reinforcement learning algorithms tell us about how the human brain implements meta-reinforcement learning“)

[Jan 13, 2020]  We have one abstract accepted to Cosyne 2020. Congratulations!

The paper is entitled
”Dynamic resource allocation during reinforcement learning accounts for ramping and phasic dopamine activity” - Minryung Song (PhD student).

[Dec 16, 2019] Our paper was published in Nature Communications.

The paper is entitled “Task complexity interacts with state-space uncertainty in the arbitration between model-based and model-free learning”. This is a collaborative work between our team and John O’Doherty at Caltech. (full text)

[Oct 7, 2019] Our paper, “Designing model-based and model-free reinforcement learning tasks without human guidance”, has been accepted to the NeurIPS workshop on Biological and Artificial Reinforcement Learning. Congratulations! The date of the workshop is on 13 December, 2019 at the Vancouver Convention Center. Feel free to come to meet our presenting authors (Jae Hoon Shin and Dr. Jee Hang Lee).

[June 3, 2019] We started KAIST Research Center for Neuroscience-Inspired Artificial Intelligence (CNAI).

cnai.png

Our research center aims to develop brain-inspired AI technology with human-like intelligence. It also promotes scientific and technological innovations in the field of computational/cognitive/developmental neuroscience, brain engineering, machine learning, and robotics. It will closely collaborate with world-class interdisciplinary research teams, including MIT Center for Brains, Minds, and Machines, University of Cambridge, University of Birmingham, Google DeepMind, IBM AI research, IBM TJ Watson, Caltech, University College London, and etc.

[media : Robot News, etnews, AI Times, Veritas-a, MBN, Yonhap News, Science Monitor, and etc.]

[2019 Spring @KAIST]

Lab_photo_2019_04-small.jpg

[April 19, 2019] Four abstracts have been accepted to RLDM. Congratulations!

  • Designing model-based and model-free reinforcement learning tasks without human guidance

  • Deciphering model-based and model-free reinforcement learning strategies and choices from EEG

  • Behavioral and neural evidence for intrinsic motivation effect on reinforcement learning

  • Metacognitive exploration in reinforcement learning

[Mar 21, 2019] A new paper, entitled “Neurostimulation Reveals Context-Dependent Arbitration Between Model-Based and Model-Free Reinforcement Learning”, was published in Cerebral Cortex. It is the first study that succeeds in neurally-tapping both MB and MF system. This is a collaborative work between our team at KAIST (Sang Wan Lee - co-first author), Caltech (John O’Doherty), and University of Zurich (Sebastian Weissengruber - co-first author, Christian Ruff).

[Feb 18, 2019] Sang Wan Lee (PI) received Songam Distinguished Research Award. It is awarded to two assistant professors of College of Engineering who have contributed to the development of KAIST and College of Engineering with outstanding research achievements.

[Feb 15, 2019] Yoondo Sung (Undergraduate student; graduation study) won the best Bachelor's thesis award. Congratulations!

[Feb 7, 2019] Two papers about brain-inspired AI have been published in Science Robotics and Current Opinion on Behavioral Sciences. This is a collaborative work between our team at KAIST (Dr. JeeHang Lee - the first author, Sujin An - co-author), U Cambridge (Dr. Ben Seymour - co-corresponding author), and Google DeepMind (Dr. Joel Leibo - co-author).

[Feb 2, 2019] Two papers have been accepted to the IEEE ICASSP : 1 oral (lecture session) and 1 poster. The ICASSP is “the world’s largest and most comprehensive technical conference focused on signal processing and its applications”.  Congratulations! The papers are entitled
”Phonemic-level duration control using attention alignment for natural speech synthesis” - Jungbae Park (PhD student),
”Polyphonic sound event detection using convolutional bidirectional LSTM and synthetic data-based transfer learning” - Seokwon Jung (Humelo Inc.).

[Jan 15, 2019]  We have one abstract accepted to Cosyne 2019. Congratulations!

The paper is entitled
”Evidence of behavioral and neural interaction between task complexity and state-space uncertainty during reinforcement learning” - Dongjae Kim (PhD student), Geon Young Park (MS student).

[DEC 14, 2018] Yujin Cha (PhD student) has been selected as a recipient of the smart healthcare research grant (44,000 USD equivalent for 32 months) from Daewoong foundation.

The aim of the project is to design a novel MD-AI cross-training paradigm. Congratulations!

[NOV 2, 2018] Yoondo Sung (Undergraduate student; individual study) and Yewon Kang (individual study) have separately received the outstanding poster award at the 2018 Korean Human Brain Mapping conference. Congratulations! The paper is entitled
“Effect of dpression on prefrontal meta-control of model-free and model-based reinforcement learning “ - Yoondo Sung,
”Decoupling novelty and uncertainty representation in the human brain during learning and inference” - Yewon Kang.

[2018 Spring @KAIST]

bml201804.PNG

[June 20, 2018]  We have six papers accepted to IEEE SMC 2018. Congratulations!

[April 21, 2018] Junyeol Kim (MS student) has received the outstanding paper award at the 2018 Korean Institute of Intelligent Systems conference.

The paper is entitled “Single agent model-based reinforcement learning with state-transition prediction”. Congratulations!

[Jan 18, 2018]  We have five abstracts accepted to Cosyne 2018. Congratulations!

We ranked the 3rd in terms of the number of accepted posters (link).

- Metacognitive exploration in a completely unknown state space
- Dynamic encoding of reward and latent task structures in human reinforcement learning
- Depressive model-based and model-free reinforcement learning
- Maximally separating and correlating model-based and model-free reinforcement learning
- Designing an experiment without a human experimenter

[DEC 1, 2017] A team of Sang Hyun Yi, Young Ho Kang, Jun Yeol Kim, and Yoondo Sung ranked 3rd in AI World Cup. 

The team achieved the highest score in the qualifying round. Congratulations!

[NOV 24, 2017] Haram Joo (Undergraduate student; individual study) has received the best paper award at the 2017 Korean Institute of Intelligent Systems conference.

The paper is entitled “Model-based reinforcement learning using probabilistic simulation”. Congratulations!

[NOV 1, 2017] A team of Jung Bae Park, a MS student of  lab, and Ja-Ryong Lee from CS dept. was selected as one of the six AI teams in the Korea AI music competition. (the music is available online http://visla.kr/?p=64375 ) On Nov 1st, the team's AI music composer, "Humelo", has been featured in the event "MUSIC X AI with SM Entertainment", co-organized by Korea Creative Content Agency, SM entertainment, Futureplay, and Korea national univ of arts, and Google Seoul campus. The demonstration of the team (Play with Errors) has received media coverage (JoongAng newspaper, Digital times - interview, Bloter - interview, China focus - interview, Yonhap news, Donga newspaper). Congratulations to both of them!

[NOV 3, 2017] Ji Eun Lim (Undergraduate student; graduation study) has received the best paper award at the 2017 Korean Society of Human Brain Mapping fall conference. The paper is entitled “EEG Synchrony Patterns of Autism Spectrum Disorder”. Congratulations!

[OCT 27, 2017] Su Jin An has received the best paper award at the 18th International Symposium on Advanced Intelligent Systems. The paper is entitled “Metacognitive Reinforcement Learning”. Congratulations!

[APR 24, 2017] Su Jin An has received the best paper award at the 2017 Korean Institute of Intelligent Systems conference. The paper is entitled “On the Exploration-Exploitation Dilemma using uncertainty based state space learning algorithm”. Congratulations!

[2017 Spring @KAIST]

bml201706.PNG

[MAR 1, 2017] The team of Benedetto de Martino (UCL) and Sang Wan Lee (KAIST) has been selected as a recipient of the 2016 Google Faculty Research Award in computational neuroscience (link).

[OCT 29, 2016] Two undergraduate students (Jung Bae Park - individual study, Juno Kim - graduation research) have received the best paper award at the 2016 Korean Institute of Intelligent Systems conference. The paper is entitled “Multi-agent Cognitive Policy Learning: Reinforcement Learning Through Competition”. Congratulations!

[SEP 1, 2016] Our lab has been awarded a research grant (1.1 Million USD over 3 years) from Samsung future technology foundation. The project is entitled "AI-human coevolution engine". It is a very competitive grant program; Samsung awarded only six research teams this  year in the intelligence for smart machine category. Our project aims to develop a framework that boosts human learning and inference performance through a combination of model-based fMRI/EEG techniques and various machine learning algorithms.
(press: yonhapnews, korea industry news, metro, newspim)