arXiv subjects

Yangsibo Huang

Publications and source records attributed to Yangsibo Huang.

41 records · Page 3Linked to original sources

arXiv2020

MixCon: Adjusting the Separability of Data Representations for Harder Data Recovery

To address the issue that deep neural networks (DNNs) are vulnerable to model inversion attacks, we design an objective function, which adjusts the separability of the hidden data representations, as a way to control the trade-off between data utility and vulnerability to inversion attacks. Our method is motivated by the theoretical insights of data separability in neural networking training and results on the hardness of model inversion. Empirically, by adjusting the separability of data representation, we show that there exist sweet-spots for data separability such that it is difficult to recover data during inference while maintaining data utility.

cs.LG↗

arXiv2020

TextHide: Tackling Data Privacy in Language Understanding Tasks

An unsolved challenge in distributed or federated learning is to effectively mitigate privacy risks without slowing down training or reducing accuracy. In this paper, we propose TextHide aiming at addressing this challenge for natural language understanding tasks. It requires all participants to add a simple encryption step to prevent an eavesdropping attacker from recovering private text data. Such an encryption step is efficient and only affects the task performance slightly. In addition, TextHide fits well with the popular framework of fine-tuning pre-trained language models (e.g., BERT) for any sentence or sentence-pair task. We evaluate TextHide on the GLUE benchmark, and our experiments show that TextHide can effectively defend attacks on shared gradients or representations and the averaged accuracy reduction is only $1.9\%$. We also present an analysis of the security of TextHide using a conjecture about the computational intractability of a mathematical problem. Our code is available at https://github.com/Hazelsuko07/TextHide

cs.CL↗

arXiv2020

Privacy-preserving Learning via Deep Net Pruning

This paper attempts to answer the question whether neural network pruning can be used as a tool to achieve differential privacy without losing much data utility. As a first step towards understanding the relationship between neural network pruning and differential privacy, this paper proves that pruning a given layer of the neural network is equivalent to adding a certain amount of differentially private noise to its hidden-layer activations. The paper also presents experimental results to show the practical implications of the theoretical finding and the key parameter values in a simple practical setting. These results show that neural network pruning can be a more effective alternative to adding differentially private noise for neural networks.

cs.LG↗

arXiv2019

DeepMCDose: A Deep Learning Method for Efficient Monte Carlo Beamlet Dose Calculation by Predictive Denoising in MR-Guided Radiotherapy

The next great leap toward improving treatment of cancer with radiation will require the combined use of online adaptive and magnetic resonance guided radiation therapy techniques with automatic X-ray beam orientation selection. Unfortunately, by uniting these advancements, we are met with a substantial expansion in the required dose information and consequential increase to the overall computational time imposed during radiation treatment planning, which cannot be handled by existing techniques for accelerating Monte Carlo dose calculation. We propose a deep convolutional neural network approach that unlocks new levels of acceleration and accuracy with regards to post-processed Monte Carlo dose results by relying on data-driven learned representations of low-level beamlet dose distributions instead of more limited filter-based denoising techniques that only utilize the information in a single dose input. Our method uses parallel UNET branches acting on three input channels before mixing latent understanding to produce noise-free dose predictions. Our model achieves a normalized mean absolute error of only 0.106% compared with the ground truth dose contrasting the 25.7% error of the under sampled MC dose fed into the network at prediction time. Our model's per-beamlet prediction time is ~220ms, including Monte Carlo simulation and network prediction, with substantial additional acceleration expected from batched processing and combination with existing Monte Carlo acceleration techniques. Our method shows promise toward enabling clinical practice of advanced treatment technologies.

physics.med-ph↗

arXiv2019

Deep Q Learning Driven CT Pancreas Segmentation with Geometry-Aware U-Net

Segmentation of pancreas is important for medical image analysis, yet it faces great challenges of class imbalance, background distractions and non-rigid geometrical features. To address these difficulties, we introduce a Deep Q Network(DQN) driven approach with deformable U-Net to accurately segment the pancreas by explicitly interacting with contextual information and extract anisotropic features from pancreas. The DQN based model learns a context-adaptive localization policy to produce a visually tightened and precise localization bounding box of the pancreas. Furthermore, deformable U-Net captures geometry-aware information of pancreas by learning geometrically deformable filters for feature extraction. Experiments on NIH dataset validate the effectiveness of the proposed framework in pancreas segmentation.

cs.CV↗