MambaGaze: Bidirectional Mamba with Explicit Missing Data Modeling for Cognitive Load Assessment from Eye-Gaze Tracking Data
arXiv:2605.22775v1 Announce Type: new Abstract: Real-time cognitive load assessment from eye-tracking signals could potentially enable adaptive human-centered-AI such as safety-critical applications such as driver vigilance monitoring or automated flight deck assistance, yet two challenges persist: handling frequent data missingness from blinks and tracking failures, and efficiently modeling long-range temporal dependencies. We propose MambaGaze, a framework that addresses these challenges through 1) XMD encoding, which augments raw features with observation masks and time-deltas to explicitly
The increasing sophistication of AI models and the demand for robust real-time human-AI interaction in safety-critical domains necessitate advanced cognitive assessment techniques. MambaGaze addresses key technical challenges in eye-tracking data analysis for this purpose.
This development is crucial for advancing adaptive human-centered AI by providing a reliable method for real-time cognitive load assessment, which can enhance safety and efficiency in complex operational environments.
The ability to accurately model missing eye-gaze data and long-range temporal dependencies changes the feasibility of deploying real-time cognitive load assessment in applications previously hindered by data quality issues.
- · Human-AI interaction designers
- · Automotive industry
- · Aerospace industry
- · AI agents developers
- · Systems relying on static, non-adaptive human interaction models
Improved human-machine teaming and reduced human error rates in demanding tasks.
Accelerated adoption of personalized adaptive interfaces that dynamically adjust to user states.
Ethical and privacy concerns around continuous, granular cognitive monitoring become more prominent.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG