How Can We Be So Dense? The Benefits of Using Highly Sparse Representations
This work addresses the robustness and efficiency issues in neural networks for machine learning practitioners, though it appears incremental as it builds on known biological insights.
The paper tackles the problem of dense representations in artificial networks by proposing highly sparse representations, showing that they offer significantly improved robustness and stability on MNIST and Google Speech Command datasets while maintaining competitive accuracy.
Most artificial networks today rely on dense representations, whereas biological networks rely on sparse representations. In this paper we show how sparse representations can be more robust to noise and interference, as long as the underlying dimensionality is sufficiently high. A key intuition that we develop is that the ratio of the operable volume around a sparse vector divided by the volume of the representational space decreases exponentially with dimensionality. We then analyze computationally efficient sparse networks containing both sparse weights and activations. Simulations on MNIST and the Google Speech Command Dataset show that such networks demonstrate significantly improved robustness and stability compared to dense networks, while maintaining competitive accuracy. We discuss the potential benefits of sparsity on accuracy, noise robustness, hyperparameter tuning, learning speed, computational efficiency, and power requirements.