1. 제     목 : Neural computation helps understand neural computation 

 

2. 발 표 자 : 김진섭 박사 (한국뇌연구원) 

3. 일    시 : 2016년  06월  22일(수)  10:30 ~ 12:30 

 

4. 장    소 : 경북대학교 IT-1호관 313호 

 

5. 초청교수 : 이민호 교수

 

6. 강사약력 :
      2015.11~Present Principal Investigator, Korea Brain Research Institute
      2014.08~2015.10 Postdoctoral Associate, Princeton Neuroscience Institute, Princeton University
      2010.08~2014.07 Postdoctoral Associate/Fellow, Department of Brain and Cognitive Science, MIT
      2010.03~2010.07 Visiting Postdoctoral Researcher, Department of Brain and Cognitive Science, Seoul National University, Korea
      2005.09~2010.02 Ph. D. in Physics, Seoul National University, Korea
      2003.03~2005.02 M. S. in Physics, Seoul National University, Korea
      1995.03~2000.02 B. S. in Physics, Seoul National University, Seoul, Korea 

 

7. 내용요약 : 
  
What is the origin of neural computation in animals’ brain? There are two possible theories.
In biophysical theory, a neuron is considered to have internal biomolecular mechanism that makes itself an already capable computer.
The connectionism, on the other hand, claims that it is rooted in the connections between neurons. 

My colleagues and I recently found a supporting evidence for the connectionism from a study about motion detection in the mammalian retina. The connectivity of retinal neurons was found by analyzing serial electron microscopic images with help from an artificial intelligence (AI) powered by convolutional neural network and also from EyeWire, an online community of ‘citizen neuroscientists’. 

As seen in this example, the neural computation of AI which was inspired by the findings on the neural computation of brain helps understand mysteries of the brain and mind. The new understandings, in turn, might help design improved AI architecture.