【Ideas and Viewpoints】
Donglin Wang: Brain research has started its new wave of revolution. The era for deep understanding of the human brain has just started.
Zhanyi Hu: Artificial Intelligence is in the impending revolution phase with the advancement of deep neural network and deep learning. Researchers in the field of Pattern Recognition and Image Understanding should pay special attention to this field of study.
Yi Zeng: Modeling and Simulating the Brain needs integration, analysis and understanding of multi-scale and multi-species brain data and knowledge. Data driven and hypothesis driven approaches are all very important and need to meet each other at some points for brain simulations.
Hua Han: In order to enable the High-throughput neural network reconstruction, a quantitative analysis system needs to be built. Pictures from different brain slices need to be integrated together to get the holistic view of the neuron. The morphologies of neurons (including synapses) need to be extracted and transformed to digital models which can be easily analyzed.
Recently, the EU and the United States started their Brain research projects respectively. International companies such as IBM and Google also released many studies related to brain simulations. Varied efforts are conducted to understand how the brain works from different perspectives. Brain research has started a new wave of revolution with participations from different research areas.
On October 25th, 2013, the first event of “The CASIA Young Scientists Innovation Salon” was hold at the CASIA Coffee Shop. This event focuses on the discussion of “The Future of Brain Science and Intelligence Science”. Professor Donglin Wang was invited to give a keynote talk. Professor Zhanyi Hu from the National Laboratory of Pattern Recognition (NLPR), Hua Han (Senior Engineer) from the National Engineering & Technology Research Center for ASIC Design (NECFAD), and Yi Zeng (Associate Professor) from the Neural Computation Group were also invited to have presentations on focused topics related to brain and intelligence research. Professor Bo Xu chaired this event.
Professor Donglin Wang surveyed recent advances and trends in Brain research worldwide. With the background on the advancement of Neurophysiology, Cognitive Science, and Computing Technology, several key projects were reviewed respectively, including the BRAIN project in the United States, the EU FET Human Brain Project, the IBM Compass Project, the Google Deep Learning Project, the UK SpiNNaker project. Professor Wang envisioned that the new era for the investigations on the understanding of the human brain has been started. For CASIA, this is a grand challenge and a big opportunity. CASIA should have its own strategy for integrating brain research and intelligence science.
Professor Zhanyi Hu envisioned that Artificial Intelligence is in the impending revolution phase with the advancement of deep neural network and deep learning. Researchers in the field of Pattern Recognition and Image Understanding should pay special attention to this field of study. He thinks that the big success of deep neural network and deep learning is that the multi-level structure of the neural network provides a wide space for semantic representation, while data driven hierarchical learning benefits from and utilizes this hierarchy. There are no abstract rules which can outperform the ability of describing the data compared to the data themselves. How to design appropriate deep neural networks for specific tasks is a serious direction. Professor Hu also pointed out that during the long period of revolution, and there are special regions which handles specific tasks in the human brain, such as the special region for face recognition. Hence, the investigation for deep neural network may need special design for specialized functions.
Dr. Yi Zeng started his talk by having a quick survey on historical contributions and advanced tools for understanding the human brain. He focused on the introduction of how to model and simulate the Brain based on the integration, analysis and understanding of multi-scale and multi-species brain data and knowledge. He introduced the Linked Neuron Data platform that his group had been building for this purpose (http://www.linked-neuron-data.org/). Dr. Zeng pointed out that data driven and hypothesis driven approaches are all very important and need to meet each other at some point for brain simulations. In addition, he had some demonstrations on neural activity predictions, micro-level simulations of mammalian vision tasks, such as neuronal activities simulation of orientation selection in the mouse and cat brain.
Dr. Hua Han focused on the topic of Neural Network micro-level reconstruction. He introduced that fast Neural Network micro-level reconstruction is a hypothesis for neuroscientists for a long time. There are many relevant studies in the Nature and the Science magazines this year. Compared to many other countries, the research in this topic is not widely discussed and conducted in China. A quantitative analysis system needs to be built. Pictures from different brain slices need to be integrated together to get the holistic view of the neuron. The morphologies of neurons (including synapses) need to be extracted and transformed to digital models which can be easily analyzed. Dr Han said that they have built an initial platform serving this purpose. Several reconstruction work based on the brain slices has been in progress. In the near future, they are going to investigate on Nano-scale, high-throughput, automatic, and scalable brain reconstruction methods and platforms.
What is the relationship between Brain science, Information and Communication Technologies? How to inspire new innovations based on the tight integration and fusion of these two disciplines? These great questions were extensively discussed during this event.
“Coffee Time·The CASIA Young Scientists Innovation Salon” is a new series of event for inspiring cross department, and cross discipline research. It aims at building an interactive, innovative atmosphere and strengthening the construction of the innovative culture at CASIA. This event is composed of invited talks, debates and discussions on certain innovative topics and is open for every researchers and students at CASIA. The event of “The Future of Brain Science and Intelligence Science” is the first attempt for planned series.