Our Publications

By integrating photonic sensing with fluorescence-based signal modulation, this work enables sensitive, multiplexed biomarker detection from small blood volumes. Quantum-enabled simulations performed on the IBM Quantum runtime were used to model and measure sensor performance using shot-based simulations (2,000 measurements per condition) of biomarker-dependent fluorescence quenching and photonic wavelength shifts at clinically relevant concentrations. These simulations allowed dose-response analysis, estimate of limit-of-detection and comparison with the known neonatal clinical decision thresholds.

Datasets for emotion recognition are typically annotated using either discrete or dimensional emotion frameworks. Discrete emotion labels are based on theories of basic emotion, which suggest that there are a limited number of emotion states (e.g., fear, happiness) associated with distinct expressions and physiological states [14]. In contrast, dimensional approaches quantify emotion along continuous axes, predominantly emotional arousal and valence [11]. Recent advances in large language models (LLMs) have expanded the scope of emotion recognition beyond the traditional paradigm of emotion label prediction. These models facilitate a more generative approach to emotion understanding, producing detailed, comprehensive descriptions of emotional states in natural language [27]. This shift has prompted the development of new datasets and metrics that accommodate rich natural language descriptions of emotions, allowing for greater nuance in emotion analysis.