Deep network as memory space: complexity, generalization, disentangled representation and interpretability

Abstract

By bridging deep networks and physics, the programme of geometrization of deep networks was proposed as a framework for the interpretability of deep learning systems. Following this programme we can apply two key ideas of physics, the geometrization of physics and the least action principle, on deep networks and deliver a new picture of deep networks: deep networks as memory space of information, where the capacity, robustness and efficiency of the memory are closely related with the complexity, generalization and disentanglement of deep networks. The key components of this understanding include:(1) a Fisher metric based formulation of the network complexity; (2)the least action (complexity=action) principle on deep networks and (3)the geometry built on deep network configurations. We will show how this picture will bring us a new understanding of the interpretability of deep learning systems.

11 pages, 2 figures

References in corpus (23)

• Understanding Black-box Predictions via Influence Functions
• The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks
• Building up spacetime with quantum entanglement
• GAN Dissection: Visualizing and Understanding Generative Adversarial Networks
• Towards Understanding Generalization of Deep Learning: Perspective of Loss Landscapes
• Switchbacks and the Bridge to Nowhere
• Quantum order from string-net condensations and origin of light and massless fermions
• Fisher-Rao Metric, Geometry, and Complexity of Neural Networks
• Deconstructing Lottery Tickets: Zeros, Signs, and the Supermask
• The Typical-State Paradox: Diagnosing Horizons with Complexity
• Algorithms for tensor network renormalization
• A theory of quantum gravity based on quantum computation
• Universal Statistics of Fisher Information in Deep Neural Networks: Mean Field Approach
• Frustration free gapless Hamiltonians for Matrix Product States
• Filter Pruning via Geometric Median for Deep Convolutional Neural Networks Acceleration
• Dear Qubitzers, GR=QM
• Dataset Distillation
• Geometric formulation of quantum mechanics
• How deep learning works --The geometry of deep learning
• Emergent General Relativity from Fisher Information Metric
• Geometrization of deep networks for the interpretability of deep learning systems
• Understanding over-parameterized deep networks by geometrization
• Gauge theory and twins paradox of disentangled representations