: I am broadly interested in machine learning, artificial intelligence, and data science. My current
research focuses on inference and learning algorithms for large scale probabilistic graphical models and its combination with
deep neural networks. My advisor is professor Nicholas
2016 - Now:
UT Dallas Computer Science Ph.D. candidate
AI and Machine Learning, Probabilistic Graphical Models, Deep Learning.
UT Dallas Computer Science M.S.
Achieved a C.S master degree at UT Dallas Computer Science Department.
End-to-end Stereo Matching with CRF Regularized Convolutional Neural Net
We propose a generic end-to-end framework for structured prediction tasks that combines neural networks and conditional random fields. We explain how our approach improves in key ways over existing combined approaches and demonstrate that the CRF acts as an effective regularizer. We demonstrate the superior performance of our combined approach on the stereo depth estimation task against pure deep neural network solutions: we plug existing deep neural network architectures into our framework and perform a head-to-head comparison on a variety of real and synthetic data sets.
Hao Xiong, Nicholas Ruozzi
General Purpose MRF Learning with Neural Network Potentials
In this work, we propose a generic MLE estimation procedure for MRFs whose potential
functions are modeled by neural networks.
To make learning effective in practice, we show how to leverage a highly parallelizable variational
inference method that can easily fit into popular machining learning frameworks like TensorFlow.
We demonstrate experimentally that our approach is capable of effectively modeling the data
distributions of a variety of real data sets and that it can compete effectively with other common
methods for multilabel classification and generative modeling tasks.
Hao Xiong, Nicholas Ruozzi
One-Shot Marginal MAP Inference in Markov Random Fields
We propose a novel variational inference strategy that is flexible enough to handle both
continuous and discrete random variables, efficient enough to be able to handle repeated
statistical inferences, and scalable enough, via modern GPUs, to be practical on MRFs with
hundreds of thousands of random variables. We prove that our approach overcomes weaknesses
of the current approaches and demonstrate the efficacy of our approach on both synthetic models and
Hao Xiong, Yuanzhen Guo *, Yibo Yang *, Nicholas Ruozzi
Marginal Inference in Continuous Markov Random Fields using Mixtures
We present an alternative
family of approximations that, instead of approximating the
messages, approximates the beliefs in the continuous Bethe
free energy using mixture distributions. We show that these
types of approximations can be combined with numerical
quadrature to yield algorithms with both theoretical guarantees on the quality of the approximation and
better practical performance in a variety of applications that
are challenging for current state-of-the-art methods.
Yuanzhen Guo, Hao Xiong, Nicholas Ruozzi