Computational Principles Underlying the Learning of Sensorimotor... (Lecture 2) by Daniel Wolpert скачать в хорошем качестве

Computational Principles Underlying the Learning of Sensorimotor... (Lecture 2) by Daniel Wolpert

PROGRAM ICTP-ICTS WINTER SCHOOL ON QUANTITATIVE SYSTEMS BIOLOGY (ONLINE) ORGANIZERS: Vijaykumar Krishnamurthy (ICTS-TIFR, India), Venkatesh N. Murthy (Harvard University, USA), Sharad Ramanathan (Harvard University, USA), Sanjay Sane (NCBS-TIFR, India) and Vatsala Thirumalai (NCBS-TIFR, India) DATE: 06 December 2021 to 17 December 2021 VENUE: Online How organisms sense the world and generate behaviors is an exciting question that has motivated neuroscientists over more than a century. Neural command for generating behavioral output arises from operations at multiple scales, ranging from the flip-flops of ion channels to dynamics in circuits comprising ensembles of neurons. New tools to genetically manipulate organisms, monitor and perturb neural activity, and advanced microscopy that enables large scale imaging of neurons in vivo have yielded a hitherto unprecedented quantum of data with high resolution. Quantitative approaches are needed to mine these data sets for generating testable hypotheses regarding nervous system function. This is the tenth school in the series on Quantitative Systems Biology, held alternately at Trieste and Bangalore. The school responds to the strong need, especially at the Ph.D. and postdoc level, for providing scientists with a broad exposure to quantitative problems in the study of living systems. The audience will range from Ph.D. students to young faculty, who either work in this area or plan to do so. QSB2021 is focussed on Sensori-motor control. The aim of this School is to expose students from different backgrounds to the latest research in systems neuroscience, with an emphasis on the usage of quantitative methods and theory. We will begin the workshop with a brief introduction to neuroscience, including electrical properties of neuronal membranes and single neuronal biophysics. With this foundation in place, we will delve into how circuit dynamics emerge in diverse circuits using invertebrate and vertebrate model organisms as examples. We will cover questions in population coding, variability and stochasticity, and plasticity. We will then introduce students to applications of quantitative tools to neuroscience data sets such as whole-brain imaging data sets or behavioral clustering data sets. Throughout, we will also explore how theory can contribute to a normative understanding of various phenomena, and motivate future experiments. Scientists and students from all over the world can apply for the School. Researchers from developing countries are particularly encouraged to apply. As the program will be conducted in English, participants should have an adequate working knowledge of this language. There is no registration fee. SCIENTIFIC ORGANISING COMMITTEE 1. Venkatesh N. Murthy (Harvard University, USA) 2. Sharad Ramanathan (Harvard University, USA) 3. Sanjay P. Sane (NCBS, India) 4. Vatsala Thirumalai (NCBS, India) SCIENTIFIC ADVISORY COMMITTEE 1. Vijay Balasubramaniam (University of Pennsylvania, USA) 2. Upinder Bhalla (NCBS, India) 3. Antonio Celani (ICTP, Italy) 4. Sanjay Jain (University of Delhi, India) 5. Vijaykumar Krishnamurthy (ICTS, India) (Local organizer) 6. Matteo Marsili (ICTP, Italy) 7. Mukund Thattai (NCBS, India) Previous editions of this school 2020 2019 2018 2017 2016 2015 2014 2013 2012 CONTACT US: qsb@icts.res.in PROGRAM LINK: https://www.icts.res.in/program/qsb2021 Table of Contents (powered by https://videoken.com) 0:00:00 Computational Principles Underlying the Learning of Sensorimotor Repertoires (Lecture 2) 0:01:48 Type of Motor Learning 0:02:14 Adaptation to novel dynamics 0:03:02 Robotic manipulandum and VR system 0:03:52 Motor learning of novel dynamics 0:06:52 Measuring adaptation 0:08:16 Contextual cues: static visual 0:09:27 Sequential action with variable dwell time 0:10:41 Future states 0:12:37 Consistent us. Variable Future 0:14:32 Planning vs. execution 0:16:46 Tool use 0:17:15 Control points on an object 0:17:48 Sensory cues: control points on an object 0:19:55 Repertoire learning 0:28:32 Inference 0:31:10 Memory creation, expression and updating 0:36:21 Which learning curve is proper learning? 0:37:55 Parameter learning in the COIN model 0:37:59 Spontaneous recovery (Smith et al. Plos Biol, 2006) 0:42:48 COIN model: Spontaneous and evoked recovery 0:44:48 Spontaneous and evoked recovery 0:46:21 1. Model class recovery 0:48:05 2. Cross-validated log likelihood 0:48:58 3. Model fitting and parameter recovery: COIN model 0:49:00 Spontaneous and evoked recovery 0:49:36 5. Parameters similar across paradigms 0:49:54 6. Robustness analysis 0:50:10 Overlearning potentates spontaneous recovery 0:50:52 Abrupt vs. gradual perturbations 0:52:47 Memory updating: Theory 0:53:51 Memory updating (n=24) 0:57:21 Classical paradigms as apparent learning 1:01:13 Explicit and implicit learning 1:03:19 Comparison with previous models 1:04:20 Summary 1:04:59 Q&A