Research

Our lab’s vision is to better understand neuroscience and neuro-medicine by designing specialized interpretable tools from machine learning and statistics.

Functional modeling in brain

We develop models and algorithms based on principals in statistics and machine learning to understand functions in the visual cortex of brain. To achieve this, we study large-scale datasets with a variety of modalities such as electrophysiology, calcium imaging, and electron microscopy, and gene expression.

Specific projects:
  • Models based on deep neural network are effective in predicting single neuron responses in primate visual cortex. Despite their high predictive accuracy, these models are generally difficult to interpret. This limits their applicability in characterizing neuron function. We investigate methods and algorithms to elicit interpretation and visualization of models of neurons based on deep neural networks. To achieve this, we develop frameworks based on stability, compression, and sparsity.
  • Characterizing the relationship between neural function and connectivity is an eminent question of visual sensory processing. In order to explore this relationship, we are analyzing recorded visual responses from pan-excitatory neurons within an 800X800 um region of primary visual cortex, spanning all visual layers from pia to white matter. This includes 750 2-photon and 35 3-photon calcium imaging planes spaced by ~16 um. Our goal is to examine the single-cell and population activity in primary visual cortex, and along with Electron Microscopic reconstruction from the same tissue, will serve as a valuable resource to study the functional connectome in mouse cortex.

Computational clinical neuroscience

In order to bridge our understanding of the brain to diagnosis, we build tools to study and visualize large-scale neurological clinical datasets. Our goals are to relate patient health records to observed symptoms as well as help clinicians to easily benefit from the information hidden in the data.  This involves analyzing and visualizing medical images and structured and unstructured clinical documents.

Specific projects:
  • In collaboration with Neuroimmunology and Glial Biology division at UCSF Neurology, we are developing quantitative methods to analyze multi-modal data from patients with neurological disease. The data includes one-of-a-kind measurements of MRI, genetic variants, electronic health record, etc from thousands of patients. Our goal is to develop analysis and modeling frameworks to identify scientifically meaningful patterns in patient data that are associated with the disease processes.

Interpretable Machine Learning

Guided by scientific questions from neuroscience and biomedicine, we are interested in the general problem of interpreting machine learning models. In the past decade, research in machine learning has been principally focused on the development of algorithms and models with high predictive capabilities. However, interpreting these models remains a challenge, primarily because of the large number of parameters involved. We investigate methods based on statistical principles to build more interpretable machine learning models.