Abstract:The bottom two layers of a neuromorphic architecture are designed and shown to be capable of online clustering and supervised classification. An active spiking dendrite model is used, and a single dendritic segment performs essentially the same function as a classic integrate-and-fire point neuron. A single dendrite is then composed of multiple segments and is capable of online clustering. Although this work focuses primarily on dendrite functionality, a multi-point neuron can be formed by combining multiple dendrites. To demonstrate its clustering capability, a dendrite is applied to spike sorting, an important component of brain-computer interface applications. Supervised online classification is implemented as a network composed of multiple dendrites and a simple voting mechanism. The dendrites operate independently and in parallel. The network learns in an online fashion and can adapt to macro-level changes in the input stream. Achieving brain-like capabilities, efficiencies, and adaptability will require a significantly different approach than conventional deep networks that learn via compute-intensive back propagation. The model described herein may serve as the foundation for such an approach.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to achieve online clustering and supervised classification in neuromorphic architectures. Specifically, the author designed the bottom layer of the neuromorphic architecture, using an activated dendritic model, and showed that a single dendritic segment can perform the same function as a classical integrate - and - fire point neuron. By combining multiple dendritic segments, a single dendrite can perform online clustering. In addition, by combining multiple dendrites, multi - point neurons can be formed, thereby achieving supervised online classification. ### Main problems 1. **Online clustering**: - The paper explored how to cluster patterns in real - time in the input stream. Each input pattern is assigned to the most similar cluster, and the cluster center is dynamically adjusted as the input changes. - The Sum of Absolute Differences (SAD) was used as a distance metric, and the nearest cluster center was determined by calculating the SAD between the input pattern and the cluster center. 2. **Supervised online classification**: - The paper proposed a network composed of multiple dendrites that work independently and in parallel, achieving supervised online classification through a simple voting mechanism. - Input patterns are grouped into different categories, each with a label. When the input pattern flows through the network, the network first predicts its label and then updates the synaptic weights in the dendrites according to the supervised input. ### Key technologies - **Activated dendritic model**: Each dendritic segment can be regarded as a simplified point neuron, which produces an output by weighted - sum input signals. - **Online learning**: Using the Spike Dependent Plasticity (SDP) mechanism, the synaptic weights are updated by the presence or absence of input and output spikes. - **Winner - Take - All (WTA) inhibition**: In the network, the dendritic segment with the maximum output value is selected, and the outputs of other dendritic segments are inhibited to 0. ### Goals - **Brain - inspired computing power**: Achieve brain - like computing power and adaptability, which requires a different method from traditional back - propagation - based deep networks. - **Efficient and real - time processing**: Be able to efficiently perform clustering and classification in real - time input streams, suitable for applications such as brain - machine interfaces. Through these methods, the paper aims to lay the foundation for the development of large - scale neuromorphic computing devices that can achieve cognitive functions in a brain - like manner.

Neuromorphic Online Clustering and Classification

NADOL: Neuromorphic Architecture for Spike-Driven Online Learning by Dendrites

Stochastic dendrites enable online learning in mixed-signal neuromorphic processing systems

Toward Efficient Processing and Learning with Spikes: New Approaches for Multispike Learning

Spike-driven Multi-Scale Learning with Hybrid Mechanisms of Spiking Dendrites.

Real-time classification of datasets with hardware embedded neuromorphic neural networks

Beyond spiking networks: The computational advantages of dendritic amplification and input segregation

NeuSort: An Automatic Adaptive Spike Sorting Approach with Neuromorphic Models

Leveraging dendritic properties to advance machine learning and neuro-inspired computing

Towards deep learning with segregated dendrites

Multibranch Formal Neuron: An Internally Nonlinear Learning Unit

Improving energy efficiency and classification accuracy of neuromorphic chips by learning binary synaptic crossbars

Efficient Spike-Driven Learning With Dendritic Event-Based Processing

Binary Resistive-Switching-Device-Based Electronic Synapse with Spike-Rate-Dependent Plasticity for Online Learning

Power-efficient neural network with artificial dendrites

Improving classification accuracy of feedforward neural networks for spiking neuromorphic chips

TripleBrain: An Edge Neuromorphic Architecture for High-accuracy Single-layer Spiking Neural Network with On-chip Self-organizing and Reinforcement Learning

Dendritic Neuron Model with Effective Learning Algorithms for Classification, Approximation, and Prediction

A System-on-Chip Based Hybrid Neuromorphic Compute Node Architecture for Reproducible Hyper-Real-Time Simulations of Spiking Neural Networks

A Neuromorphic-Hardware Oriented Bio-Plausible Online-Learning Spiking Neural Network Model.