Skip to content

Latest commit

 

History

History
68 lines (44 loc) · 2.32 KB

README.md

File metadata and controls

68 lines (44 loc) · 2.32 KB
Raw

系统与计算神经科学作业

本项目是基于RSNN的模拟小鼠决策任务的程序。程序模拟了小鼠的感知、工作记忆和决策实验。实验设定小鼠在丁字形通道内，先经过带有提示信息的通道，再经过没有信息的通道，最终路口处决定左转或右转。通过奖励信息对小鼠行为进行训练。本程序通过计算建模对此实验进行模拟。搭建RSNN循环脉冲神经网络模拟小鼠大脑活动，通过监督训练模拟小鼠的学习行为，最终使得网络具有感知、工作记忆和决策能力。

详细报告内容见 小组项目报告

Requirements

  • Python >=3.6
  • Pytorch =1.12.1
  • braincog =0.2.7.15
  • matplotlib =3.6.2

Setting things up

Using Pip

To install the most recent stable release from the GitHub repository

pip install git+https://github.com/jdy18/Crossing.git

Getting started

To run a script of the RSNN with trainging method of e-prop, issue

python main.py

Environment

Mouse experience setting

Result

Activation of each neuron when 5~1000 samples have been trained

Activation of each neuron when 5~1000 samples have been trained

Changes of membrane potential of each neuron with time at the beginning of training

Changes of membrane potential of each neuron with time at the end of training

All simulations run on Ubuntu 16.04 LTS with Intel(R) Xeon(R) CPU E5-2687W v3 @ 3.10GHz, 128Gb RAM @ 2133MHz, and two GeForce GTX TITAN X (GM200) GPUs. Python 3.6 is used in all cases. Clock time was recorded for each simulation run.

Contributors

  • 马国庆 (学号：202218020428003 email )
  • 张予涵 (学号：202218014628014 email)
  • 江德扬 (学号：202218020428001 email)
  • 魏雅轩 (学号：202218020415011 email)
  • 林楚儿 (学号：202218020428002 email）