Model- and Layer-Level Closed-Loop Forecasting with SSM-Based Spiking Neural Networks

Key points

• Spiking neural networks (SNNs) constitute a biologically inspired class of deep SSMs, where computing blocks communicate through spikes, discrete single-bit events rather than continuous activation functions.1
• We explore two closed-loop forecasting strategies, in which the network's own generated values are fed back into the network as if they were observed past values: (i) model-level and (ii) layer-level closed-loop forecasting.1
• We evaluate these approaches on the ETT-small datasets, a standard benchmark for time-series forecasting, under both spike-based and continuous activation functions.1
• These findings highlight the trade-off between continuous and low-bit quantized activations, and demonstrate that the proposed Deep SSM architectures provide a competitive framework for forecasting.1

Sources (1)

• [1]Model- and Layer-Level Closed-Loop Forecasting with SSM-Based Spiking Neural Networks
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 20, 04:39 PM

  Deep State-Space Models (Deep SSMs) have shown promising performance and computational benefits in processing of time-series data with complex temporal dependencies.

  Spiking neural networks (SNNs) constitute a biologically inspired class of deep SSMs, where computing blocks communicate through spikes, discrete single-bit events rather than continuous activation functions.

Extractive summary: sentences quoted from the sources.