Sensory Information Processing Laboratory

Our laboratory conducts research at the intersection of sensory neuroscience, machine learning, and biomedical engineering, with a particular emphasis on hearing and vision. A central goal of the laboratory is to understand how sensory systems encode and process information, and to use these principles to develop clinically useful methods for diagnosis, screening, and rehabilitation.

One major research theme focuses on the application of machine learning and Bayesian methods to hearing science and audiology. The laboratory develops data-driven approaches for rapid hearing aid fitting and auditory assessment, with the aim of reducing clinical testing time while improving accuracy and personalization. This work combines psychophysics, statistical modelling, and adaptive testing strategies to estimate auditory characteristics efficiently from limited measurements. The laboratory is also interested in auditory brainstem responses (ABR) as a method for probing “hidden hearing loss”, namely hearing deficits that may not appear in conventional audiological tests but nevertheless impair communication, particularly in noisy environments.

A second major area of research concerns machine learning approaches for glaucoma screening and visual field testing. Conventional visual field assessment can be time-consuming and fatiguing for patients, limiting its effectiveness for large-scale screening. The laboratory develops multidimensional Bayesian and machine learning frameworks that exploit the statistical structure of visual field data to accelerate testing while maintaining diagnostic reliability. These approaches integrate computational modelling with ophthalmological data to create adaptive screening paradigms capable of identifying glaucomatous deficits using fewer measurements. The broader objective is to improve access to early detection technologies for visual disorders while reducing patient burden.

In parallel with these applied projects, the laboratory conducts fundamental research on neural information processing in sensory systems. This work examines how single neurons and neural populations encode external stimuli, with particular interest in adaptation, efficient coding, variability, and low-dimensional representations of neural activity. The laboratory develops theoretical and computational models that aim to explain sensory coding using minimal assumptions and few free parameters, drawing inspiration from statistical physics, signal processing, and dynamical systems theory. These studies seek to connect the behavior of individual neurons with systems-level sensory function.

By combining theoretical neuroscience, machine learning, and clinical applications, the laboratory aims to better understand sensory information processing while contributing to practical technologies for hearing and vision assessment.