Compute-in-Memory Implementation of State Space Models for Event Sequence Processing

2025/11/17 by Xiaoyu Zhang, Zhang, Xiaoyu, Mingtao Hu +10 · 2 citations
Computer Science · Engineering · #Advanced Memory and Neural Computing #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #FOS: Electrical engineering #Ferroelectric and Negative Capacitance Devices #Machine Learning (cs.LG) #Neural Networks and Reservoir Computing #Signal Processing (eess.SP) #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.2511.13912

openalex publication_date 2025/11/17 · openalex created_date 2025/11/20 · openalex updated_date 2026/07/28

Abstract

State space models (SSMs) have recently emerged as a powerful framework for long sequence processing, outperforming traditional methods on diverse benchmarks. Fundamentally, SSMs can generalize both recurrent and convolutional networks and have been shown to even capture key functions of biological systems. Here we report an approach to implement SSMs in energy-efficient compute-in-memory (CIM) hardware to achieve real-time, event-driven processing. Our work re-parameterizes the model to function with real-valued coefficients and shared decay constants, reducing the complexity of model mapping onto practical hardware systems. By leveraging device dynamics and diagonalized state transition parameters, the state evolution can be natively implemented in crossbar-based CIM systems combined with memristors exhibiting short-term memory effects. Through this algorithm and hardware co-design, we show the proposed system offers both high accuracy and high energy efficiency while supporting fully asynchronous processing for event-based vision and audio tasks.

Citations

• Neuromorphic Principles for Efficient Large Language Models on Intel Loihi 2
• Scalable MatMul-free Language Modeling
• Sovereign Byte Firewall: Comparing Transformer and Mamba-2 Backbones for Tokenization-Free Zero-Day Network Anomaly Detection
• Scalable Event-by-event Processing of Neuromorphic Sensory Signals With Deep State-Space Models
• Jamba: A Hybrid Transformer-Mamba Language Model
• Gated Linear Attention Transformers with Hardware-Efficient Training
• Mamba: Linear-Time Sequence Modeling with Selective State Spaces
• Convolutional State Space Models for Long-Range Spatiotemporal Modeling
• GET: Group Event Transformer for Event-Based Vision
• Resurrecting Recurrent Neural Networks for Long Sequences
• SpikeGPT: Generative Pre-trained Language Model with Spiking Neural Networks
• Spikformer: When Spiking Neural Network Meets Transformer
• Simplified State Space Layers for Sequence Modeling
• Efficient recurrent architectures through activity sparsity and sparse back-propagation through time
• Efficiently Modeling Long Sequences with Structured State Spaces
• Training Spiking Neural Networks Using Lessons From Deep Learning
• Temporal Binary Representation for Event-Based Action Recognition
• HiPPO: Recurrent Memory with Optimal Polynomial Projections
• Spintronics for neuromorphic computing
• CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features
• End-to-End Learning of Representations for Asynchronous Event-Based Data
• Quantization and Training of Neural Networks for Efficient\n Integer-Arithmetic-Only Inference
• Attention Is All You Need
• Deep Residual Learning for Image Recognition
• BinaryConnect: Training Deep Neural Networks with binary weights during propagations
• Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Coding
• Networks of spiking neurons: The third generation of neural network models

Cited by

• Neuromorphic Object Detection: An In-Depth Study and Future Directions
• Low-latency Event-based Object Detection with Spatially-Sparse Linear Attention

Related

• IMSSA: Deploying modern state-space models on memristive in-memory compute hardware
• A Diagonal Structured State Space Model on Loihi 2 for Efficient Streaming Sequence Processing
• Systolic Array-based Accelerator for Structured State-Space Models
• Simple Hardware-Efficient Long Convolutions for Sequence Modeling
• Resistive memory-based zero-shot liquid state machine for multimodal event data learning