“2010세계 뇌 주간”을 맞이하여 뇌과학의 중요성을 일반인이 이해하고 뇌에 대한 관심을 증진시키고자 “뇌를 쉽게 알려줍니다”라는 주제로 행사를 개최하게 되었습니다.
뇌주간 행사(Brain Awareness Week)는 1992년 미국에서 시작된 이래 현재 60여개국에서 매년 3월 셋째주에 동시에 진행되는 세계적인 행사입니다. 한국에서도 2002년부터 행사를 진행하고 있으며 올해는 3월 13(토)부터 20(토) 사이에 전국적으로 개최됩니다.
대전지역은 3월 13일(토) KAIST에서 아래와 같이 개최됩니다.
일시: 2010년 3월 13일(토) 오후 10:00-13:30
장소: KAIST LG Semicon Hall(N24) 1층 강당
주최: KAIST 뇌과학연구센터 / 전기및전자공학과
주제: 뇌과학, 뇌공학, 그리고 미래사회
10:00-10:40 지상 최대의 쇼: 뇌와 유전자의 격투기 (KAIST 김대수)
10:40-11:20 디지털 세대 청소년의 뇌는 무엇이 다른가? (고려대학교 한종혜)
11:20-12:00 뇌과학과 철학: 마지막 남은 미지의 세계 (KAIST 김대식)
12:10-12:50 인공두뇌: 로봇 태권브이의 두뇌 (KAIST 이수영)
12:50-13:30 뇌파로 움직이는 장난감 (데모 및 실습) (NeuroSky 임종진)
뇌과학연구센터 세미나 개최
일시: 2009년 8월 26일(수) 오후 2:00-3:30
장소: KAIST 정문술빌딩 219호
제목: Auditory Temporal Assimilation: A Classifier and Information Analysis of
연사: Hiroshige Takeichi (Research Scientist, RIKEN BSI, Japan)
A part of the data from an event-related potential (ERP) experiment on auditory temporal assimilation was reanalyzed by constructing Gaussian Naive Bayes Classifiers. In auditory
temporal assimilation, two neighboring time intervals marked by three successive tone
bursts appear to have the same duration as an illusion if the first time interval and the
second time interval satisfy a relationship.
The classifiers could discriminate the subject's task, which was judgment of the
equivalence between the two intervals, at an accuracy of 88 to 96% as well as their
subjective judgments to the physically equivalent stimulus at an accuracy of 82 to 86%
from individual average waveforms. Chernoff mutual information provided more consistent interpretations compared with classification errors as to the selection of the component most strongly associated with the perceptual judgment. This may provide us with a robust neurodecoding scheme.
(2.826MB, DN: 1329)
2008년 호암 공학상 기념 강연
일시: 2008년 6월 4일(수) 오후 4시~5시 30분
장소: KAIST 정문술빌딩 드림홀
강연제목: From the Brain to the Computer and Back Again
연사 : Hyunjune Sebastian Seung
Professor of Computational Neuroscience, Brain & Cognitive Sciences Dept., MIT
Professor of Physics, Physics Dept., MIT
Investigator, Howard Hughes Medical Institute
Engineers have long dreamed of building a machine with the capabilities of the human mind. Today's digital computers are adroit at long division or chess, surpassing even the best humans. But they falter at other mental tasks, like recognizing objects from images, or communicating via natural language. Artificial intelligence (AI) is still more dream than reality.
The brain is the organ that gives rise to the mind. Naturally, some engineers have studied the brain, hoping to discover principles that could be used for building intelligent machines. In one approach to AI, artificial neural networks are constructed or simulated so as to mimic the biological neural networks inside the brain. This bio-mimetic approach to AI is a bridge between neuroscience and computer science, and dates back to the very dawn of the computer age.
One important feature of human intelligence is the ability to learn. In recent years, machine learning has become a major focus in the field of AI. An example of machine learning is my work on nonnegative matrix factorization (NMF), done in collaboration with Daniel Lee. A computer can use NMF to extract a relatively small number of features from a large database of vectors representing images, sounds, or other kinds of information. This is done by optimizing an objective function under nonnegativity constraints. In this purely mathematical formulation, there is no reference to the brain. However, our interest in NMF originally grew out of the bio-mimetic approach to AI. If NMF is implemented by an artificial neural network, then the nonnegativity constraints can be interpreted as basic constraints on neural computation in the brain: the activities of neurons are never negative, and the efficacy of signal transmission at a synapse never changes sign.
I have just described research driven by a central question: what can neuroscience do for computer science? But computer science has made astonishing progress over the past half century, even if the dream of artificial intelligence has not been realized. Computers have transformed human culture, whereas the advances of neuroscience mostly remain within the confines of the laboratory. Computer science is like the younger brother who has grown up to be taller and stronger. Today it makes sense to turn the original question around: what can computer science do for neuroscience?
One answer is that computers will be important for acquiring and analyzing the immense amounts of data that will be generated by neuroscientists in the years to come. In particular, three-dimensional nanoscale imaging of brain structure is starting to produce gargantuan databases. For example, imaging a cubic millimeter at 20 nanometer resolution yields more than 100 terabytes of data. In principle, a great deal can be learned from such images about the structure of biological neural networks. Through image analysis, it should be possible to identify all synapses, and determine the pair of neurons corresponding to each synapse. This would yield a map of all the connections in a biological neural network, its "connectome."
In the future, connectomes could be important for understanding how biological neural networks function, how the brain wires itself up during development, and how some mental disorders could be due to miswiring of the brain. In practice, finding connectomes is extremely tedious if image analysis is done manually by humans. Today my laboratory works on computer automation of image analysis. Our goal is to create methods that will be useful for finding connectomes. Image analysis is a difficult and important problem in artificial intelligence. We use artificial neural networks to analyze images of biological neural networks. The recursive quality of this strategy shows that computer science and neuroscience can each benefit from the other. The relationship between these two fields continues to be fruitful in new and exciting ways.
ⓒ Copyright 1999~ TECHNOTE-TOP / TECHNOTE.INC,
Copyright (c) Brain Science Research Center. All right reserved