InfoNet: Neural Estimation of Mutual Information without Test-Time Optimization

Estimating mutual correlations between random variables or data streams is essential for intelligent behavior and decision-making. As a fundamental quantity for measuring statistical relationships, mutual information has been extensively studied and utilized for its generality and equitability. However, existing methods often lack the efficiency needed for real-time applications, such as test-time optimization of a neural network, or the differentiability required for end-to-end learning, like histograms. We introduce a neural network called InfoNet, which directly outputs mutual information estimations of data streams by leveraging the attention mechanism and the computational efficiency of deep learning infrastructures. By maximizing a dual formulation of mutual information through large-scale simulated training, our approach circumvents time-consuming test-time optimization and offers generalization ability. We evaluate the effectiveness and generalization of our proposed mutual information estimation scheme on various families of distributions and applications. Our results demonstrate that InfoNet and its training process provide a graceful efficiency-accuracy trade-off and order-preserving properties. We will make the code and models available as a comprehensive toolbox to facilitate studies in different fields requiring real-time mutual information estimation.

• We proposed InfoNet, the first pre-learned mutual information model, which could output the estimation through only one feed-forward pass.
• We conduct large amount of experiments to validate InfoNet's efficient speed, strong generalization and scalable ability.
• InfoNet has many potential applications in the future since it satisfies the need for large-scale real-time mutual information estimation. One work using InfoNet is MaxMI: A Maximal Mutual Information Criterion for Manipulation Concept Discovery (ECCV 2024 Oral).