Emily Hsiang

WORK HISTORY

EDUCATION

ABOUT EMILY HSIANG

Deep Learning & Neural Network Development hybrid attention-based deep neural networks in PyTorch, leveraging Docker, GitHub, and HPC clusters to predict neural activity and integrate diverse datasets. Designed biologically constrained architectures to ensure physiological relevance. Built and optimized end-to-end pipelines for large-scale foundation model training and deployment in generative AI. Prioritized model interpretability and the disentanglement of underlying factors in low-data Signal Processing & Machine Learning and implemented signal extraction algorithms (e.g, optical flow) and machine learning models (e.g, MDS, KPCA) to analyze retinal bipolar cell responses to naturalistic stimuli. Employed semi-supervised learning for partially labeled data. Applied inferential statistics—bootstrap and nonparametric Visual System Modeling & Information Theory optical modeling and information theory to study neuron encoding strategies in the human visual system. Conducted Fourier and receptive field neuronal modeling for population-level Image Processing & 3D Data Analysis image processing pipelines (MATLAB) for 3D volume registration, denoising, and de novo, morphology-aware functional segmentation algorithms. Extracted neural features from high-dimensional imaging Two-Photon Microscopy & Calcium Imaging a two-photon microscope for ex vivo calcium imaging in transgenic mouse models. Investigated subcellular sensory processing in the early visual Human Cognitive Neuroscience Stanford, analyzed intracranial recordings and deep brain stimulation data to explore electrical coupling between brain regions. In Taiwan, utilized EEG and non-invasive brain stimulation (tDCS, TMS) to study neural interactions in attention tasks.