UCLA Team Releases OpenSLA: Unifying Sensor Time Series, Language and Actions
yang_yuzhe · x · 2026-10-08
Yuzhe Yang's UCLA team introduced Sensor-Language-Action (SLA) models and the open-source OpenSLA framework, pushing sensor AI beyond perception toward deciding what to do next — and why.
- Unified framework: one model handles sensor time series, natural language, and actions, with a shared hierarchy covering action necessity, category, and fine-grained label.
- State understanding: generates captions recovering rates, ranges, and waveform patterns directly from sensor history.
- Evidence grounding: every action prediction comes with the sensor findings and patient context supporting it.
- Zero-shot transfer: distinguishes unseen actions from no-action and evaluates on the external MIMIC-IV clinical cohort without fine-tuning.
- Future state: frozen representations support readouts of glucose dynamics over the next two hours.
Healthcare is the main domain (ED records, ECG, respiration, HR, CGM). Paper, code, and Hugging Face weights are all public.
More from Research
- EngramEdit: decoupled factual knowledge edits in LLMs via conditional memory — ModalityDance · 2026-10-08
- Last Translation Benchmark adds North, Mistral Large 4, Qwen 3.8; open for coauthorship — zouharvi · 2026-10-08
- Chrome's new DecisionModel API reverse-engineered: prompts, limits and engine tests — dejanseo · 2026-10-08
- New Study Decodes Neural Codes of Speech Comprehension Over Time — abenitezburraco · 2026-10-08
- Mathematicians: AI reshapes problem solving but won't displace mathematicians — tianyin_xu · 2026-10-08
- Isomorphic Labs aims to 'solve all disease' with generative AI built beyond AlphaFold — ns123abc · 2026-10-08