A Compressed Sensing Based Decomposition of Electrodermal Activity Signals.

The measurement and analysis of electrodermal activity (EDA) offers applications in diverse areas ranging from market research to seizure detection and to human stress analysis. Unfortunately, the analysis of EDA signals is made difficult by the superposition of numerous components that can obscure the signal information related to a user's response to a stimulus. We show how simple preprocessing followed by a novel compressed sensing based decomposition can mitigate the effects of the undesired noise components and help reveal the underlying physiological signal. The proposed framework allows for decomposition of EDA signals with provable bounds on the recovery of user responses. We test our procedure on both synthetic and real-world EDA signals from wearable sensors and demonstrate that our approach allows for more accurate recovery of user responses as compared with the existing techniques.

Detecting divisions of the autonomic nervous system using wearables.

The ability to assess a user's emotional reaction from biometrics has applications in personalization, recommendation, and enhancing user experiences, among other areas. Unfortunately, understanding the connection between biometric signals and user reactions has previously focused on black box techniques that are opaque to the underlying physiology of the user. In this paper, we explore a novel user study connecting biometric reaction to external stimuli and changes in the user's autonomic nervous system. Specifically, we focus on two competing responses, namely the sympathetic and parasympathetic nervous system, and how differing activations are related to different user responses. Our experiments demonstrate how prior psychophysiological research distinguishing this activation can be replicated using biometric data collected from wearables. The insights from this work have applications in better understanding emotional state from biometric sensors.

Statistical detection and imaging of objects hidden in turbid media using ballistic photons.

We exploit recent advances in active high-resolution imaging through scattering media with ballistic photons. We derive the fundamental limits on the accuracy of the estimated parameters of a mathematical model that describes such an imaging scenario and compare the performance of ballistic and conventional imaging systems. This model is later used to derive optimal single-pixel statistical tests for detecting objects hidden in turbid media. To improve the detection rate of the aforementioned single-pixel detectors, we develop a multiscale algorithm based on the generalized likelihood ratio test framework. Moreover, considering the effect of diffraction, we derive a lower bound on the achievable spatial resolution of the proposed imaging systems. Furthermore, we present the first experimental ballistic scanner that directly takes advantage of novel adaptive sampling and reconstruction techniques.