Общеуниверситетский факультатив "Моделирование когнитивных процессов". Читает профессор НИУ ВШЭ Dr. Дж. Макиннес

Общеуниверситетский факультатив "Моделирование когнитивных процессов". Читает профессор НИУ ВШЭ Dr. Дж. Макиннес
Факультатив проходит по адресу Мясницкая, 11. Запись полная.

Computational Modelling

by W. Joseph MacInnes, PhD, associate professor in the department of psychology, HSE

 

 

Computational modelling is the practice of simulating cognitive processes and neuronal function using machine learning and other computer algorithms.

Psychology experiments have always been driven by theories of how the brain works.  Current theories help us design the best experiments, and the results improve our best theories.  But if a theory stands and falls based on how well it fits the data from our experiments, how can we measure that across many diverse experiments? Computational modelling has allowed us to express our theories in strong, unambiguous form and test them mathematically against data from psychology and neuroscience experiments. This non-credit workshop will cover a number of machine learning algorithms used in cognitive modelling as well as discuss existing computational models that have been developed in many areas of psychology.

A back ground in Mathematics/statistics/programming would be helpful, but it is not required. The workshop will meet weekly over the course of the term. Sessions will include both theoretical discussions of modelling research as well as hands on sessions of related machine learning techniques. Journal club style discussions will be held on published computational models such as Itti and Koch’s salience model of low level vision, Ratcliffe’s diffusion model of decision processes and Dehaene’s model of consciousness. In addition, we will have hands on sessions where students learn tools to develop their own models using Bayesian networks, Hidden Markov models, neural networks and diffusion models. Finally, students will also be introduced to IBM ‘Watson’ as an example of applied modelling in an industry setting. ‘Watson’ is the computer program that won on the Jeopardy game-show, but contains algorithms which can be applied to many modelling projects including language, comprehension and interaction. By the end of the workshop, the entire group will work on a project to submit to computational modelling contest (The IBM great mind challenge, a poster to a computational modelling conference, the Loebner prize, Robo-cup contest are all possibilities).