Abstract:Analyzing a visual scene by inferring the configuration of a generative model is widely considered the most flexible and generalizable approach to scene understanding. Yet, one major problem is the computational challenge of the inference procedure, involving a combinatorial search across object identities and poses. Here we propose a neuromorphic solution exploiting three key concepts: (1) a computational framework based on Vector Symbolic Architectures (VSA) with complex-valued vectors; (2) the design of Hierarchical Resonator Networks (HRN) to factorize the non-commutative transforms translation and rotation in visual scenes; (3) the design of a multi-compartment spiking phasor neuron model for implementing complex-valued resonator networks on neuromorphic hardware. The VSA framework uses vector binding operations to form a generative image model in which binding acts as the equivariant operation for geometric transformations. A scene can, therefore, be described as a sum of vector products, which can then be efficiently factorized by a resonator network to infer objects and their poses. The HRN features a partitioned architecture in which vector binding is equivariant for horizontal and vertical translation within one partition and for rotation and scaling within the other partition. The spiking neuron model allows mapping the resonator network onto efficient and low-power neuromorphic hardware. Our approach is demonstrated on synthetic scenes composed of simple 2D shapes undergoing rigid geometric transformations and color changes. A companion paper demonstrates the same approach in real-world application scenarios for machine vision and robotics.

Holographic Graph Neuron: a Bio-Inspired Architecture for Pattern Processing

Pattern Recognition Computation in a Spiking Neural Network with Temporal Encoding and Learning

A Brain-Inspired Spiking Neural Network Model with Temporal Encoding and Learning

Holographic Neural Architectures

A Temporal Encoding Method Based on Expansion Representation

A Spiking Neural Network System for Robust Sequence Recognition

Rapid Feedforward Computation by Temporal Encoding and Learning with Spiking Neurons

A Neuro-Cognitive System and Its Application in Robotics

Residual and Attentional Architectures for Vector-Symbols

Learning Real-World Stimuli By Single-Spike Coding And Tempotron Rule

Graphene-Based Artificial Dendrites for Bio-Inspired Learning in Spiking Neuromorphic Systems

Learning Through Structure: Towards Deep Neuromorphic Knowledge Graph Embeddings

Hyperdimensional Computing with Spiking-Phasor Neurons

Homogeneous Spiking Neuromorphic System for Real-World Pattern Recognition

Neuromorphic Visual Scene Understanding with Resonator Networks

Hyperseed: Unsupervised Learning with Vector Symbolic Architectures

The Cooperative Network Architecture: Learning Structured Networks as Representation of Sensory Patterns

A Method for the Symbolic Representation of Neurons

A structural representing and learning model based on biological neural mechanism