Data-Efficient Neural Mutual Information Estimation for Capturing Brain-to-Brain Communication
Technical Report,01 Oct 2017,30 Apr 2019
SRI International Princeton United States
Pagination or Media Count:
Measuring Mutual Information MI between high-dimensional, continuous, random variables from observed samples has wide theoretical and practical applications. Traditional MI methods, capable of capturing MI between low-dimensional signals, fall short when dimensionality increases and are not scalable. Existing neural approaches search for a d-dimensional neural network that maximizes a variational lower bound for mutual information estimation however, this requires Od log d observed samples to prevent the neural network from overfitting. For practical mutual information estimation in real world applications, data is not always available at a surplus, especially in cases where acquisition of the data is prohibitively expensive, for example in fMRI analysis. This effort introduces a scalable, data-efficient mutual information estimator. BY coupling a learning-based view of the MI lower bound with meta-learning, NeuralMI achieves high-confidence estimations irrespective of network size and with improved accuracy at practical dataset sizes. The effectiveness has been demonstrated on synthetic benchmarks as well as a real world application of fMRI inter-subject correlation analysis.
- Anatomy and Physiology