The human brain is the most complex system in the world, but it is essentially an information processing system, and the information is generally sent out by the way ofactionpotentiation. Whole Brain-mind Machine(WBM) platform provides the biological brain structure information provided by the collected multi-dimensional and multi-scale biological brain data, including the brain structure magnetic resonance data and functional magnetic resonance data, and uses traditional biophysical models to simulateasingle neuron and the interaction between neurons, develop a data assimilation method, estimate hyperparameters, build a WBM model with no less than 10 million neurons, andconstructa digital brain finally.

The WBM platform aims to make breakthroughs in basic brain science research and basic artificial intelligence algorithms, and provides open hardware for research in the fields of neuroscience/brain science, brain disease diagnosis and treatment technology, human brain enhancement technology, and brain-inspiredintelligence technology. The computing hardware and open source software tools will provide a practical platform for the study of brain science and the development of a new generation of brain-inspired intelligence.

Annual construction progress

At present, the WBM platform has 5 full-time researchers, 2 senior algorithm engineers, 2 research assistants, and 7PHD students. At the same time, it has long-term stable corporation with the Micro-Nano System Center of the School of Information, the School of Computer Science and Technology, and the School of Data Science.

In 2021, the platform hasbuilt a WBM model based on spiking neural network and brain imaging data, and estimate model parameters by developing a mesoscopic-scale data assimilation method. Based on the real structure and fMRI data,theWBM model including 90 brain regions and 10 million neurons is established, andtheWBM model including more than 20,000 voxels and 100 million neurons at the voxel level, And more than 20,000 voxels multiplied by six layers of functional column-level WBM models. The results showed that the data assimilation achieved good results, and the results of the model were in good agreement with the experimentally measured brain data. Based on the 75% similarity of the voxel version of the resting state simulation, a series of task experiments including information flow and split-brain were performed. These results further confirm that building a WBM model based on spiking neural network and brain magnetic resonance data is feasible.