AI & ML
DS3M: Deep structured state space models for nonlinear system identification on modern hardware
arXiv cs.SY0 views
arXiv:2611.25274v1 Announce Type: cross Abstract: Neural state-space identification learns dynamical models from measured inputs and outputs. SUBNET combines this task with an encoder that estimates initial states from short measurement histories. Its continuous-time version requires many successive neural-network evaluations during numerical integration, resulting in slow training speed which is a major limitation for practical use cases. We propose deep structured state space models (DS3M), which combine compact nonlinear dynamical components in parallel and in series. Custom GPU kernels implemented with Triton reduce the launch and memory overhead of their sequential evaluation. We explain this construction, describe a port-Hamiltonian variant, and present various benchmark comparisons of accuracy and training time.
Source:https://arxiv.org/abs/2611.25274PreviousSmall-Gain Analysis of Robust Event-Triggered Extremum Seeking for Scalar MapsNextThe Ces\`aro Value Iteration
Related posts
ChatGPT Helped Me Complain About ChatGPT. The Irony Wasn't Lost on Me.
0HackerNoon
Inside North Korea’s Crypto Theft Machine: Hackers, Launderers, and Billions in Stolen Funds
0HackerNoon
EC-Council Releases ADG 2.0 and Offers Free AI Governance Crosswalks to Governments Worldwide
0HackerNoon
The Boutique PR Firm Is Having Its Moment
0HackerNoon
Blockchain Was Raised in the 2008 Financial Crisis. Now It’s Respawning It
0HackerNoon
Why Anthropic's Claude Turned to Blackmail in AI Safety Tests
0HackerNoon