arXiv subjects

Zhi Zhang

Publications and source records attributed to Zhi Zhang.

At least 145 records · Page 8Linked to original sources

arXiv2020

Improving Semantic Segmentation via Self-Training

Deep learning usually achieves the best results with complete supervision. In the case of semantic segmentation, this means that large amounts of pixelwise annotations are required to learn accurate models. In this paper, we show that we can obtain state-of-the-art results using a semi-supervised approach, specifically a self-training paradigm. We first train a teacher model on labeled data, and then generate pseudo labels on a large set of unlabeled data. Our robust training framework can digest human-annotated and pseudo labels jointly and achieve top performances on Cityscapes, CamVid and KITTI datasets while requiring significantly less supervision. We also demonstrate the effectiveness of self-training on a challenging cross-domain generalization task, outperforming conventional finetuning method by a large margin. Lastly, to alleviate the computational burden caused by the large amount of pseudo labels, we propose a fast training schedule to accelerate the training of segmentation models by up to 2x without performance degradation.

cs.CV↗

arXiv2020

GluonCV and GluonNLP: Deep Learning in Computer Vision and Natural Language Processing

We present GluonCV and GluonNLP, the deep learning toolkits for computer vision and natural language processing based on Apache MXNet (incubating). These toolkits provide state-of-the-art pre-trained models, training scripts, and training logs, to facilitate rapid prototyping and promote reproducible research. We also provide modular APIs with flexible building blocks to enable efficient customization. Leveraging the MXNet ecosystem, the deep learning models in GluonCV and GluonNLP can be deployed onto a variety of platforms with different programming languages. The Apache 2.0 license has been adopted by GluonCV and GluonNLP to allow for software distribution, modification, and usage.

cs.LG↗

arXiv2019

Design and Evaluation of a Multi-Domain Trojan Detection Method on Deep Neural Networks

This work corroborates a run-time Trojan detection method exploiting STRong Intentional Perturbation of inputs, is a multi-domain Trojan detection defence across Vision, Text and Audio domains---thus termed as STRIP-ViTA. Specifically, STRIP-ViTA is the first confirmed Trojan detection method that is demonstratively independent of both the task domain and model architectures. We have extensively evaluated the performance of STRIP-ViTA over: i) CIFAR10 and GTSRB datasets using 2D CNNs, and a public third party Trojaned model for vision tasks; ii) IMDB and consumer complaint datasets using both LSTM and 1D CNNs for text tasks; and speech command dataset using both 1D CNNs and 2D CNNs for audio tasks. Experimental results based on 28 tested Trojaned models demonstrate that STRIP-ViTA performs well across all nine architectures and five datasets. In general, STRIP-ViTA can effectively detect Trojan inputs with small false acceptance rate (FAR) with an acceptable preset false rejection rate (FRR). In particular, for vision tasks, we can always achieve a 0% FRR and FAR. By setting FRR to be 3%, average FAR of 1.1% and 3.55% are achieved for text and audio tasks, respectively. Moreover, we have evaluated and shown the effectiveness of STRIP-ViTA against a number of advanced backdoor attacks whilst other state-of-the-art methods lose effectiveness in front of one or all of these advanced backdoor attacks.

cs.CR↗

arXiv2019

PiBooster: A Light-Weight Approach to Performance Improvements in Page Table Management for Paravirtual Virtual-Machines

In paravirtualization, the page table management components of the guest operating systems are properly patched for the security guarantees of the hypervisor. However, none of them pay enough attentions to the performance improvements, which results in two noticeable performance issues. First, such security patches exacerbate the problem that the execution paths of the guest page table (de)allocations become extremely long, which would consequently increase the latencies of process creations and exits. Second, the patches introduce many additional IOTLB flushes, leading to extra IOTLB misses, and the misses would have negative impacts on I/O performance of all peripheral devices. In this paper, we propose PiBooster, a novel lightweight approach for improving the performance in page table management. First, PiBooster shortens the execution paths of the page table (de)allocations by the PiBooster cache, which maintains dedicated buffers for serving page table (de)allocations. Second, PiBooster eliminates the additional IOTLB misses with a fine-grained validation scheme, which performs page table and DMA validations separately, instead of doing both together. We implement a prototype on Xen with Linux as the guest kernel. We do small modifications on Xen (166 SLoC) and Linux kernel (350 SLoC). We evaluate the I/O performance in both micro and macro ways. The micro experiment results indicate that PiBooster is able to completely eliminate the additional IOTLB flushes in the workload-stable environments, and effectively reduces (de)allocation time of the page table by 47% on average. The macro benchmarks show that the latencies of the process creations and exits are expectedly reduced by 16% on average. Moreover, the SPECINT,lmbench and netperf results indicate that PiBooster has no negative performance impacts on CPU computation, network I/O, and disk I/O.

cs.OS↗

arXiv2019

CATTmew: Defeating Software-only Physical Kernel Isolation

All the state-of-the-art rowhammer attacks can break the MMU-enforced inter-domain isolation because the physical memory owned by each domain is adjacent to each other. To mitigate these attacks, physical domain isolation, introduced by CATT, physically separates each domain by dividing the physical memory into multiple partitions and keeping each partition occupied by only one domain. CATT implemented physical kernel isolation as the first generic and practical software-only defense to protect kernel from being rowhammered as kernel is one of the most appealing targets. In this paper, we develop a novel exploit that could effectively defeat the physical kernel isolation and gain both root and kernel privileges. Our exploit can work without exhausting the page cache or the system memory, or relying on the information of the virtual-to-physical address mapping. The exploit is motivated by our key observation that the modern OSes have double-owned kernel buffers (e.g., video buffers and SCSI Generic buffers) owned concurrently by the kernel and user domains. The existence of such buffers invalidates the physical kernel isolation and makes the rowhammer-based attack possible again. Existing conspicuous rowhammer attacks achieving the root/kernel privilege escalation exhaust the page cache or even the whole system memory. Instead, we propose a new technique, named memory ambush. It is able to place the hammerable double-owned kernel buffers physically adjacent to the target objects (e.g., page tables) with only a small amount of memory. As a result, our exploit is stealthier and has fewer memory footprints. We also replace the inefficient rowhammer algorithm that blindly picks up addresses to hammer with an efficient one. Our algorithm selects suitable addresses based on an existing timing channel.

cs.CR↗

arXiv2019

Integrating independent and centralized multi-agent reinforcement learning for traffic signal network optimization

Traffic congestion in metropolitan areas is a world-wide problem that can be ameliorated by traffic lights that respond dynamically to real-time conditions. Recent studies applying deep reinforcement learning (RL) to optimize single traffic lights have shown significant improvement over conventional control. However, optimization of global traffic condition over a large road network fundamentally is a cooperative multi-agent control problem, for which single-agent RL is not suitable due to environment non-stationarity and infeasibility of optimizing over an exponential joint-action space. Motivated by these challenges, we propose QCOMBO, a simple yet effective multi-agent reinforcement learning (MARL) algorithm that combines the advantages of independent and centralized learning. We ensure scalability by selecting actions from individually optimized utility functions, which are shaped to maximize global performance via a novel consistency regularization loss between individual utility and a global action-value function. Experiments on diverse road topologies and traffic flow conditions in the SUMO traffic simulator show competitive performance of QCOMBO versus recent state-of-the-art MARL algorithms. We further show that policies trained on small sub-networks can effectively generalize to larger networks under different traffic flow conditions, providing empirical evidence for the suitability of MARL for intelligent traffic control.

cs.LG↗

arXiv2019

Dynamic Mini-batch SGD for Elastic Distributed Training: Learning in the Limbo of Resources

With an increasing demand for training powers for deep learning algorithms and the rapid growth of computation resources in data centers, it is desirable to dynamically schedule different distributed deep learning tasks to maximize resource utilization and reduce cost. In this process, different tasks may receive varying numbers of machines at different time, a setting we call elastic distributed training. Despite the recent successes in large mini-batch distributed training, these methods are rarely tested in elastic distributed training environments and suffer degraded performance in our experiments, when we adjust the learning rate linearly immediately with respect to the batch size. One difficulty we observe is that the noise in the stochastic momentum estimation is accumulated over time and will have delayed effects when the batch size changes. We therefore propose to smoothly adjust the learning rate over time to alleviate the influence of the noisy momentum estimation. Our experiments on image classification, object detection and semantic segmentation have demonstrated that our proposed Dynamic SGD method achieves stabilized performance when varying the number of GPUs from 8 to 128. We also provide theoretical understanding on the optimality of linear learning rate scheduling and the effects of stochastic momentum.

cs.LG↗

arXiv2019

Just-in-Time Dynamic-Batching

Batching is an essential technique to improve computation efficiency in deep learning frameworks. While batch processing for models with static feed-forward computation graphs is straightforward to implement, batching for dynamic computation graphs such as syntax trees or social network graphs is challenging due to variable computation graph structure across samples. Through simulation and analysis of a Tree-LSTM model, we show the key trade-off between graph analysis time and batching effectiveness in dynamic batching. Based on this finding, we propose a dynamic batching method as an extension to MXNet Gluon's just-in-time compilation (JIT) framework. We show empirically that our method yields up to 6.25 times speed-up on a common dynamic workload, a tree-LSTM model for the semantic relatedness task.

cs.DC↗

arXiv2019

Bag of Freebies for Training Object Detection Neural Networks

Training heuristics greatly improve various image classification model accuracies~\cite{he2018bag}. Object detection models, however, have more complex neural network structures and optimization targets. The training strategies and pipelines dramatically vary among different models. In this works, we explore training tweaks that apply to various models including Faster R-CNN and YOLOv3. These tweaks do not change the model architectures, therefore, the inference costs remain the same. Our empirical results demonstrate that, however, these freebies can improve up to 5% absolute precision compared to state-of-the-art baselines.

cs.CV↗

arXiv2018

Bag of Tricks for Image Classification with Convolutional Neural Networks

Much of the recent progress made in image classification research can be credited to training procedure refinements, such as changes in data augmentations and optimization methods. In the literature, however, most refinements are either briefly mentioned as implementation details or only visible in source code. In this paper, we will examine a collection of such refinements and empirically evaluate their impact on the final model accuracy through ablation study. We will show that, by combining these refinements together, we are able to improve various CNN models significantly. For example, we raise ResNet-50's top-1 validation accuracy from 75.3% to 79.29% on ImageNet. We will also demonstrate that improvement on image classification accuracy leads to better transfer learning performance in other application domains such as object detection and semantic segmentation.

cs.CV↗

arXiv2018

KASR: A Reliable and Practical Approach to Attack Surface Reduction of Commodity OS Kernels

Commodity OS kernels have broad attack surfaces due to the large code base and the numerous features such as device drivers. For a real-world use case (e.g., an Apache Server), many kernel services are unused and only a small amount of kernel code is used. Within the used code, a certain part is invoked only at runtime while the rest are executed at startup and/or shutdown phases in the kernel's lifetime run. In this paper, we propose a reliable and practical system, named KASR, which transparently reduces attack surfaces of commodity OS kernels at runtime without requiring their source code. The KASR system, residing in a trusted hypervisor, achieves the attack surface reduction through a two-step approach: (1) reliably depriving unused code of executable permissions, and (2) transparently segmenting used code and selectively activating them. We implement a prototype of KASR on Xen-4.8.2 hypervisor and evaluate its security effectiveness on Linux kernel-4.4.0-87-generic. Our evaluation shows that KASR reduces the kernel attack surface by 64% and trims off 40% of CVE vulnerabilities. Besides, KASR successfully detects and blocks all 6 real-world kernel rootkits. We measure its performance overhead with three benchmark tools (i.e., SPECINT, httperf and bonnie++). The experimental results indicate that KASR imposes less than 1% performance overhead (compared to an unmodified Xen hypervisor) on all the benchmarks.

cs.CR↗

arXiv2018

Progressive Neural Networks for Image Classification

The inference structures and computational complexity of existing deep neural networks, once trained, are fixed and remain the same for all test images. However, in practice, it is highly desirable to establish a progressive structure for deep neural networks which is able to adapt its inference process and complexity for images with different visual recognition complexity. In this work, we develop a multi-stage progressive structure with integrated confidence analysis and decision policy learning for deep neural networks. This new framework consists of a set of network units to be activated in a sequential manner with progressively increased complexity and visual recognition power. Our extensive experimental results on the CIFAR-10 and ImageNet datasets demonstrate that the proposed progressive deep neural network is able to obtain more than 10 fold complexity scalability while achieving the state-of-the-art performance using a single network model satisfying different complexity-accuracy requirements.

cs.CV↗

arXiv2018

Tunable magnetic and transport properties of Mn3Ga thin films on Ta/Ru seedlayer

Hexagonal D019-type Mn3Z alloys that possess large anomalous and topological-like Hall effects have attracted much attention due to their great potential in the antiferromagnetic spintronic devices. Here, we report the preparation of Mn3Ga film in both tetragonal and hexagonal phases with a tuned Ta/Ru seed layer on the thermally oxidized Si substrate. A large coercivity together with a large anomalous Hall resistivity is found in the Ta-only sample with mixed tetragonal phase. By increasing the thickness of Ru layer, the tetragonal phase gradually disappears and a relatively pure hexagonal phase is obtained in the Ta(5)/Ru(30) buffered sample. Further magnetic and transport measurements revealed that the anomalous Hall conductivity nearly vanishes in the pure hexagonal sample, while an abnormal asymmetric hump structure emerges in the low field region. The extracted additional Hall term is robust in a large temperature range and presents a sign reversal above 200K. The abnormal Hall properties are proposed to be closely related with the frustrated spin structure of D019 Mn3Ga.

cond-mat.mtrl-sci↗

arXiv2017

Knowledge Projection for Deep Neural Networks

While deeper and wider neural networks are actively pushing the performance limits of various computer vision and machine learning tasks, they often require large sets of labeled data for effective training and suffer from extremely high computational complexity. In this paper, we will develop a new framework for training deep neural networks on datasets with limited labeled samples using cross-network knowledge projection which is able to improve the network performance while reducing the overall computational complexity significantly. Specifically, a large pre-trained teacher network is used to observe samples from the training data. A projection matrix is learned to project this teacher-level knowledge and its visual representations from an intermediate layer of the teacher network to an intermediate layer of a thinner and faster student network to guide and regulate its training process. Both the intermediate layers from the teacher network and the injection layers from the student network are adaptively selected during training by evaluating a joint loss function in an iterative manner. This knowledge projection framework allows us to use crucial knowledge learned by large networks to guide the training of thinner student networks, avoiding over-fitting, achieving better network performance, and significantly reducing the complexity. Extensive experimental results on benchmark datasets have demonstrated that our proposed knowledge projection approach outperforms existing methods, improving accuracy by up to 4% while reducing network complexity by 4 to 10 times, which is very attractive for practical applications of deep neural networks.

cs.CV↗

arXiv2017

Knowledge-Guided Deep Fractal Neural Networks for Human Pose Estimation

Human pose estimation using deep neural networks aims to map input images with large variations into multiple body keypoints which must satisfy a set of geometric constraints and inter-dependency imposed by the human body model. This is a very challenging nonlinear manifold learning process in a very high dimensional feature space. We believe that the deep neural network, which is inherently an algebraic computation system, is not the most effecient way to capture highly sophisticated human knowledge, for example those highly coupled geometric characteristics and interdependence between keypoints in human poses. In this work, we propose to explore how external knowledge can be effectively represented and injected into the deep neural networks to guide its training process using learned projections that impose proper prior. Specifically, we use the stacked hourglass design and inception-resnet module to construct a fractal network to regress human pose images into heatmaps with no explicit graphical modeling. We encode external knowledge with visual features which are able to characterize the constraints of human body models and evaluate the fitness of intermediate network output. We then inject these external features into the neural network using a projection matrix learned using an auxiliary cost function. The effectiveness of the proposed inception-resnet module and the benefit in guided learning with knowledge projection is evaluated on two widely used benchmarks. Our approach achieves state-of-the-art performance on both datasets.

cs.CV↗

arXiv2016

Joint Audio-Video Fingerprint Media Retrieval Using Rate-Coverage Optimization

In this work, we propose a joint audio-video fingerprint Automatic Content Recognition (ACR) technology for media retrieval. The problem is focused on how to balance the query accuracy and the size of fingerprint, and how to allocate the bits of the fingerprint to video frames and audio frames to achieve the best query accuracy. By constructing a novel concept called Coverage, which is highly correlated to the query accuracy, we are able to form a rate-coverage model to translate the original problem into an optimization problem that can be resolved by dynamic programming. To the best of our knowledge, this is the first work that uses joint audio-video fingerprint ACR technology for media retrieval with a theoretical problem formulation. Experimental results indicate that compared to reference algorithms, the proposed method has up to 25% query accuracy improvement while using 60% overall bit-rates, and 25% bit-rate reduction while achieving 85% accuracy, and it significantly outperforms the solution with single audio or video source fingerprint.

cs.IR↗

arXiv2016

Spatially Supervised Recurrent Convolutional Neural Networks for Visual Object Tracking

In this paper, we develop a new approach of spatially supervised recurrent convolutional neural networks for visual object tracking. Our recurrent convolutional network exploits the history of locations as well as the distinctive visual features learned by the deep neural networks. Inspired by recent bounding box regression methods for object detection, we study the regression capability of Long Short-Term Memory (LSTM) in the temporal domain, and propose to concatenate high-level visual features produced by convolutional networks with region information. In contrast to existing deep learning based trackers that use binary classification for region candidates, we use regression for direct prediction of the tracking locations both at the convolutional layer and at the recurrent unit. Our extensive experimental results and performance comparison with state-of-the-art tracking methods on challenging benchmark video tracking datasets shows that our tracker is more accurate and robust while maintaining low computational cost. For most test video sequences, our method achieves the best tracking performance, often outperforms the second best by a large margin.

cs.CV↗

arXiv2016

A Capture-gated Fast Neutron Detection Method

To address the problem of the shortage of neutron detectors used in radiation portal monitors (RPMs), caused by the 3He supply crisis, research on a cadmium-based capture-gated fast neutron detector is presented in this paper. The detector is composed of many 1 cm * 1 cm * 20 cm plastic scintillator cuboids covered by 0.1 mm thick film of cadmium. The detector uses cadmium to absorb thermal neutrons and produce capture gamma-rays to indicate the detection of neutrons, and uses plastic scintillator to moderate neutrons and register gamma-rays. This design removes the volume competing relationship in traditional 3He counter-based fast neutron detectors, which hinders enhancement of the neutron detection efficiency. Detection efficiency of 21.66 +- 1.22% has been achieved with a 40.4 cm * 40.4 cm * 20 cm overall detector volume. This detector can measure both neutrons and gamma-rays simultaneously. A small detector (20.2 cm * 20.2 cm * 20 cm) demonstrated a 3.3 % false alarm rate for a 252Cf source with a neutron yield of 1841 n/s from 50 cm away within 15 seconds measurement time. It also demonstrated a very low (< 0.06%) false alarm rate for a 3.21 * 105 Bq 137Cs source. This detector offers a potential single-detector replacement for both neutron and the gamma-ray detectors in RPM systems.

physics.ins-det↗