Shangqian Gao
University of Pittsburgh
pruning
multimodality
weight-sharing
representation learning for vision
deep neural network algorithms
learning on the edge & model compression
deep generative models & autoencoders
deep neural architectures
svd
deep learning deep neural architectures
lora
efficient language models
efficient large language models
cross-modal content generation
cross-modal pretraining
4
presentations
6
number of views
SHORT BIO
I am now a fifth-year Ph.D. student in ECE, Swanson School of Engineering, University of Pittsburgh, supervised by Prof. Heng Huang. Prior to that, I obtained my M.S. degree in Computer System Engineering at Northeastern University in 2017, and I obtained my B.S. degree in Electronic Engineering at Xidian University in 2015.
My research interests includes Efficient Deep Learning, Cross-modal Learning, Reinforcement Learning with Policy Gradient Methods and Zeroth-order (gradient-free) Optimization methods.
Presentations
Unlocking Memorization in Large Language Models with Dynamic Soft Prompting
Zhepeng Wang and 8 other authors
Adaptive Rank Selections for Low-Rank Approximation of Language Models
Shangqian Gao and 4 other authors
Compressing Image-to-Image Translation GANs Using Local Density Structures on Their Learned Manifold
Alireza Ganjdanesh and 3 other authors
EffConv: Efficient Learning of Kernel Sizes for Convolution Layers of CNNs
Alireza Ganjdanesh and 2 other authors