Sparse Communication via Mixed Distributions

Abstract

Neural networks and other machine learning models compute continuous representations, while humans communicate mostly through discrete symbols. Reconciling these two forms of communication is desirable for generating human-readable interpretations or learning discrete latent variable models, while maintaining end-to-end differentiability. Some existing approaches (such as the Gumbel-Softmax transformation) build continuous relaxations that are discrete approximations in the zero-temperature limit, while others (such as sparsemax transformations and the Hard Concrete distribution) produce discrete/continuous hybrids. In this paper, we build rigorous theoretical foundations for these hybrids, which we call "mixed random variables." Our starting point is a new "direct sum" base measure defined on the face lattice of the probability simplex. From this measure, we introduce new entropy and Kullback-Leibler divergence functions that subsume the discrete and differential cases and have interpretations in terms of code optimality. Our framework suggests two strategies for representing and sampling mixed random variables, an extrinsic ("sample-and-project") and an intrinsic one (based on face stratification). We experiment with both approaches on an emergent communication benchmark and on modeling MNIST and Fashion-MNIST data with variational auto-encoders with mixed latent variables.

Accepted for oral presentation at ICLR 2022

References in corpus (15)

• The NumPy array: a structure for efficient numerical computation
• Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms
• Categorical Reparameterization with Gumbel-Softmax
• Learning to Communicate with Deep Multi-Agent Reinforcement Learning
• The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables
• Implicit Reparameterization Gradients
• SparseMAP: Differentiable Sparse Structured Inference
• The continuous Bernoulli: fixing a pervasive error in variational autoencoders
• Learning with Differentiable Perturbed Optimizers
• Differentiable Perturb-and-Parse: Semi-Supervised Parsing with a Structured Variational Autoencoder
• DVAE++: Discrete Variational Autoencoders with Overlapping Transformations
• Emergent Multi-Agent Communication in the Deep Learning Era
• On Controllable Sparse Alternatives to Softmax
• The continuous categorical: a novel simplex-valued exponential family
• Efficient Marginalization of Discrete and Structured Latent Variables via Sparsity