Spiking Brains

Study the effect of memory on current response (to behavioral tasks) in mice using methods from computational neuroscience and machine learning.

Installation

Clone the repository.

git clone https://github.com/theairbend3r/spiking-brains.git

Install the packages.

Using Conda.

conda env create -f spiking-brains.yml

Using Pip.

pip install requirements.txt

Content

The modules reside in the package ./app.

Following are the notebooks that use function from ./app/ to perform analysis.

1. Exploratory Analysis
2. Behaviour Analysis
3. Neurons Analysis
4. Phenomena Analysis
5. Machine Learning Modelling

Experiment and Analysis

Goal

• Study the effect of memory on current response (to behavioral tasks) in mice using machine learning.

Hypothesis

• Previous responses to visual stimulus, by the mouse, may affect its present response.

Dataset

• A subset of the Steinmetz dataset (Steinmetz et al, 2019).
• It contains 39 sessions from 10 mice.
• The mice were shown 2 images and had to determine which image had the highest contrast.

Method

• Train a logistic regression model to predict the mouse's response given the following input variables for current timestamp - 1.
  • Feedback type
  • Feedback time
  • Reward time
  • Response type
  • Contrast left
  • Contrast right
• Tune the model and use 8-fold cross validation to gauge accuracy.
• Plot the confusion matrix to compare the actual mouse response vs the model's response.
• Analyse the beta-weights per input variable to see its effect on the response.

Results

• See 05_modelling.ipynb notebook.

Meta

Akshaj Verma – @theairbend3r.

Distributed under the GNU GPL-V3 license. See LICENSE for more information.

https://github.com/theairbend3r/spiking-brains