New Phase State Space Models enable parallel training of spiking neural networks

Researchers have introduced Phase State Space Models (PSSMs) as a novel framework for training spiking neural networks in parallel. This approach offers a new interpretation of resonate-and-fire (R&F) neural networks, making them compatible with both real and spiking inputs, and parallel or recurrent execution. The PSSM framework also establishes connections to hyperdimensional computing and retains biologically realistic features, with an implementation demonstrated that incorporates short-time Fourier transforms, recurrent memory, and attentional capabilities. AI

IMPACT Introduces a new parallel training method for spiking neural networks, potentially improving efficiency and biological realism in AI research.

RANK_REASON The cluster contains a new academic paper detailing a novel method for training neural networks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.NE (Neural & Evolutionary) →

• arXiv
• Hyperdimensional (HD) computing
• Phase State Space Models
• Recurrent Neural Networks
• Resonate-and-fire (R&F) neural networks
• short-time Fourier transform
• Spiking networks for Bayesian inference and choice
• State Space Models
• Wilkie Olin-Ammentorp

• paper
• infra

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →