Search Results for author: Ziming Zhang

Found 50 papers, 8 papers with code

SBO-RNN: Reformulating Recurrent Neural Networks via Stochastic Bilevel Optimization

no code implementations NeurIPS 2021 Ziming Zhang, Yun Yue, Guojun Wu, Yanhua Li, Haichong Zhang

In this paper we consider the training stability of recurrent neural networks (RNNs) and propose a family of RNNs, namely SBO-RNN, that can be formulated using stochastic bilevel optimization (SBO).

Bilevel Optimization

CoFi: Coarse-to-Fine ICP for LiDAR Localization in an Efficient Long-lasting Point Cloud Map

no code implementations19 Oct 2021 Yecheng Lyu, Xinming Huang, Ziming Zhang

In addition, we propose a map based LiDAR localization algorithm that extracts semantic feature points from the LiDAR frames and apply CoFi to estimate the pose on an efficient point cloud map.

Semantic Segmentation

Toward Realistic Backdoor Injection Attacks on DNNs using Rowhammer

no code implementations14 Oct 2021 M. Caner Tol, Saad Islam, Berk Sunar, Ziming Zhang

To this end, we first investigate the viability of backdoor injection attacks in real-life deployments of DNNs on hardware and address such practical issues in hardware implementation from a novel optimization perspective.

Learning Lightweight Neural Networks via Channel-Split Recurrent Convolution

no code implementations29 Sep 2021 Guojun Wu, Yun Yue, Yanhua Li, Ziming Zhang

Lightweight neural networks refer to deep networks with small numbers of parameters, which are allowed to be implemented in resource-limited hardware such as embedded systems.

An Optimization Perspective on Realizing Backdoor Injection Attacks on Deep Neural Networks in Hardware

no code implementations29 Sep 2021 M. Caner Tol, Saad Islam, Berk Sunar, Ziming Zhang

Recent works focus on software simulation of backdoor injection during the inference phase by modifying network weights, which we find often unrealistic in practice due to the hardware restriction such as bit allocation in memory.

Stabilized Likelihood-based Imitation Learning via Denoising Continuous Normalizing Flow

no code implementations29 Sep 2021 Xin Zhang, Yanhua Li, Ziming Zhang, Christopher Brinton, Zhenming Liu, Zhi-Li Zhang, Hui Lu, Zhihong Tian

State-of-the-art imitation learning (IL) approaches, e. g, GAIL, apply adversarial training to minimize the discrepancy between expert and learner behaviors, which is prone to unstable training and mode collapse.

Denoising Imitation Learning

Auto-Encoding Inverse Reinforcement Learning

no code implementations29 Sep 2021 Kaifeng Zhang, Rui Zhao, Ziming Zhang, Yang Gao

Reinforcement learning (RL) provides a powerful framework for decision-making, but its application in practice often requires a carefully designed reward function.

Decision Making Imitation Learning +1

Revisiting 2D Convolutional Neural Networks for Graph-based Applications

no code implementations23 May 2021 Yecheng Lyu, Xinming Huang, Ziming Zhang

Graph convolutional networks (GCNs) are widely used in graph-based applications such as graph classification and segmentation.

Graph Classification Point Cloud Segmentation

Deep Lucas-Kanade Homography for Multimodal Image Alignment

1 code implementation CVPR 2021 Yiming Zhao, Xinming Huang, Ziming Zhang

With those properties, directly updating the Lucas-Kanade algorithm on our feature maps will precisely align image pairs with large appearance changes.

A Surface Geometry Model for LiDAR Depth Completion

1 code implementation17 Apr 2021 Yiming Zhao, Lin Bai, Ziming Zhang, Xinming Huang

Therefore, it is assumed those pixels share the same surface with the nearest LiDAR point, and their respective depth can be estimated as the nearest LiDAR depth value plus a residual error.

Depth Completion Self-Supervised Learning

Training Deep Neural Networks via Branch-and-Bound

1 code implementation5 Apr 2021 Yuanwei Wu, Ziming Zhang, Guanghui Wang

In this paper, we propose BPGrad, a novel approximate algorithm for deep nueral network training, based on adaptive estimates of feasible region via branch-and-bound.

Object Recognition Stochastic Optimization

EllipsoidNet: Ellipsoid Representation for Point Cloud Classification and Segmentation

no code implementations3 Mar 2021 Yecheng Lyu, Xinming Huang, Ziming Zhang

In recent years, point cloud representation in 2D space has attracted increasing research interest since it exposes the local geometry features in a 2D space.

Classification General Classification +1

Self-Supervised Geometric Features Discovery via Interpretable Attention for Vehicle Re-Identification and Beyond

no code implementations ICCV 2021 Ming Li, Xinming Huang, Ziming Zhang

To learn distinguishable patterns, most of recent works in vehicle re-identification (ReID) struggled to redevelop official benchmarks to provide various supervisions, which requires prohibitive human labors.

Representation Learning Self-Supervised Learning +1

f-GAIL: Learning f-Divergence for Generative Adversarial Imitation Learning

no code implementations NeurIPS 2020 Xin Zhang, Yanhua Li, Ziming Zhang, Zhi-Li Zhang

This naturally gives rise to the following question: Given a set of expert demonstrations, which divergence can recover the expert policy more accurately with higher data efficiency?

Imitation Learning

Discovering Discriminative Geometric Features with Self-Supervised Attention for Vehicle Re-Identification and Beyond

no code implementations19 Oct 2020 Ming Li, Xinming Huang, Ziming Zhang

Specifically, we implement an end-to-end trainable deep network architecture consisting of three branches: (1) a global branch as backbone for image feature extraction, (2) an attentional branch for producing attention masks, and (3) a self-supervised branch for regularizing the attention learning with rotated images to locate geometric features.

Representation Learning Self-Supervised Learning +2

$f$-GAIL: Learning $f$-Divergence for Generative Adversarial Imitation Learning

1 code implementation2 Oct 2020 Xin Zhang, Yanhua Li, Ziming Zhang, Zhi-Li Zhang

This naturally gives rise to the following question: Given a set of expert demonstrations, which divergence can recover the expert policy more accurately with higher data efficiency?

Imitation Learning

LodoNet: A Deep Neural Network with 2D Keypoint Matchingfor 3D LiDAR Odometry Estimation

no code implementations1 Sep 2020 Ce Zheng, Yecheng Lyu, Ming Li, Ziming Zhang

Deep learning based LiDAR odometry (LO) estimation attracts increasing research interests in the field of autonomous driving and robotics.

Autonomous Driving

Automatic Building and Labeling of HD Maps with Deep Learning

no code implementations1 Jun 2020 Mahdi Elhousni, Yecheng Lyu, Ziming Zhang, Xinming Huang

This approach speeds up the process of building and labeling HD maps, which can make meaningful contribution to the deployment of autonomous vehicle.

Autonomous Driving

RNNs Incrementally Evolving on an Equilibrium Manifold: A Panacea for Vanishing and Exploding Gradients?

no code implementations ICLR 2020 Anil Kag, Ziming Zhang, Venkatesh Saligrama

Recurrent neural networks (RNNs) are particularly well-suited for modeling long-term dependencies in sequential data, but are notoriously hard to train because the error backpropagated in time either vanishes or explodes at an exponential rate.

Learning to Segment 3D Point Clouds in 2D Image Space

1 code implementation CVPR 2020 Yecheng Lyu, Xinming Huang, Ziming Zhang

In contrast to the literature where local patterns in 3D point clouds are captured by customized convolutional operators, in this paper we study the problem of how to effectively and efficiently project such point clouds into a 2D image space so that traditional 2D convolutional neural networks (CNNs) such as U-Net can be applied for segmentation.

3D Part Segmentation graph construction

Self-Orthogonality Module: A Network Architecture Plug-in for Learning Orthogonal Filters

no code implementations5 Jan 2020 Ziming Zhang, Wenchi Ma, Yuanwei Wu, Guanghui Wang

In this paper, we investigate the empirical impact of orthogonality regularization (OR) in deep learning, either solo or collaboratively.

Graph-Preserving Grid Layout: A Simple Graph Drawing Method for Graph Classification using CNNs

no code implementations26 Sep 2019 Yecheng Lyu, Xinming Huang, Ziming Zhang

Graph convolutional networks (GCNs) suffer from the irregularity of graphs, while more widely-used convolutional neural networks (CNNs) benefit from regular grids.

Data Augmentation General Classification +1

White-Box Adversarial Defense via Self-Supervised Data Estimation

1 code implementation13 Sep 2019 Zudi Lin, Hanspeter Pfister, Ziming Zhang

In this paper, we study the problem of how to defend classifiers against adversarial attacks that fool the classifiers using subtly modified input data.

Adversarial Defense Self-Supervised Learning

Towards Learning Affine-Invariant Representations via Data-Efficient CNNs

no code implementations31 Aug 2019 Xenju Xu, Guanghui Wang, Alan Sullivan, Ziming Zhang

In this paper we propose integrating a priori knowledge into both design and training of convolutional neural networks (CNNs) to learn object representations that are invariant to affine transformations (i. e., translation, scale, rotation).

Translation

Unsupervised Deep Feature Transfer for Low Resolution Image Classification

no code implementations27 Aug 2019 Yuanwei Wu, Ziming Zhang, Guanghui Wang

We use pre-trained convenet to extract features for both high- and low-resolution images, and then feed them into a two-layer feature transfer network for knowledge transfer.

Classification General Classification +2

RNNs Evolving on an Equilibrium Manifold: A Panacea for Vanishing and Exploding Gradients?

no code implementations22 Aug 2019 Anil Kag, Ziming Zhang, Venkatesh Saligrama

Recurrent neural networks (RNNs) are particularly well-suited for modeling long-term dependencies in sequential data, but are notoriously hard to train because the error backpropagated in time either vanishes or explodes at an exponential rate.

Verification of Very Low-Resolution Faces Using An Identity-Preserving Deep Face Super-Resolution Network

no code implementations26 Mar 2019 Esra Ataer-Cansizoglu, Michael Jones, Ziming Zhang, Alan Sullivan

Face super-resolution methods usually aim at producing visually appealing results rather than preserving distinctive features for further face identification.

Face Identification Face Recognition +2

Time-Delay Momentum: A Regularization Perspective on the Convergence and Generalization of Stochastic Momentum for Deep Learning

no code implementations2 Mar 2019 Ziming Zhang, Wenju Xu, Alan Sullivan

In this paper we study the problem of convergence and generalization error bound of stochastic momentum for deep learning from the perspective of regularization.

Equilibrated Recurrent Neural Network: Neuronal Time-Delayed Self-Feedback Improves Accuracy and Stability

no code implementations2 Mar 2019 Ziming Zhang, Anil Kag, Alan Sullivan, Venkatesh Saligrama

We show that such self-feedback helps stabilize the hidden state transitions leading to fast convergence during training while efficiently learning discriminative latent features that result in state-of-the-art results on several benchmark datasets at test-time.

Deformable Part Networks

no code implementations22 May 2018 Ziming Zhang, Rongmei Lin, Alan Sullivan

In this paper we propose novel Deformable Part Networks (DPNs) to learn {\em pose-invariant} representations for 2D object recognition.

Object Recognition

LMKL-Net: A Fast Localized Multiple Kernel Learning Solver via Deep Neural Networks

no code implementations22 May 2018 Ziming Zhang

In contrast to previous works, as a learning principle we propose {\em parameterizing} both the gating function for learning kernel combination weights and the multiclass classifier in LMKL using an attentional network (AN) and a multilayer perceptron (MLP), respectively.

Convergent Block Coordinate Descent for Training Tikhonov Regularized Deep Neural Networks

no code implementations NeurIPS 2017 Ziming Zhang, Matthew Brand

By lifting the ReLU function into a higher dimensional space, we develop a smooth multi-convex formulation for training feed-forward deep neural networks (DNNs).

BPGrad: Towards Global Optimality in Deep Learning via Branch and Pruning

no code implementations CVPR 2018 Ziming Zhang, Yuanwei Wu, Guanghui Wang

Understanding the global optimality in deep learning (DL) has been attracting more and more attention recently.

Object Recognition

Learning Joint Feature Adaptation for Zero-Shot Recognition

no code implementations23 Nov 2016 Ziming Zhang, Venkatesh Saligrama

In this paper we propose a novel framework of learning data-dependent feature transforms for scoring similarity between an arbitrary pair of source and target data instances to account for the wide variability in target domain.

Zero-Shot Learning

Real-Time Visual Tracking: Promoting the Robustness of Correlation Filter Learning

no code implementations29 Aug 2016 Yao Sui, Ziming Zhang, Guanghui Wang, Yafei Tang, Li Zhang

By exploiting the anisotropy of the filter response, three sparsity related loss functions are proposed to alleviate the overfitting issue of previous methods and improve the overall tracking performance.

Real-Time Visual Tracking

Zero-Shot Learning via Joint Latent Similarity Embedding

no code implementations CVPR 2016 Ziming Zhang, Venkatesh Saligrama

It takes an arbitrary pair of source and target domain instances as input and predicts whether or not they come from the same class, i. e. whether there is a match.

Dictionary Learning Zero-Shot Learning

Sequential Optimization for Efficient High-Quality Object Proposal Generation

no code implementations14 Nov 2015 Ziming Zhang, Yun Liu, Xi Chen, Yanjun Zhu, Ming-Ming Cheng, Venkatesh Saligrama, Philip H. S. Torr

We propose a novel object proposal algorithm, BING++, which inherits the virtue of good computational efficiency of BING but significantly improves its proposal localization quality.

Object Proposal Generation

Efficient Training of Very Deep Neural Networks for Supervised Hashing

no code implementations CVPR 2016 Ziming Zhang, Yu-Ting Chen, Venkatesh Saligrama

In this paper, we propose training very deep neural networks (DNNs) for supervised learning of hash codes.

Group Membership Prediction

no code implementations ICCV 2015 Ziming Zhang, Yu-Ting Chen, Venkatesh Saligrama

In this context we propose a novel probability model and introduce latent {\em view-specific} and {\em view-shared} random variables to jointly account for the view-specific appearance and cross-view similarities among data instances.

Person Re-Identification

A Novel Visual Word Co-occurrence Model for Person Re-identification

no code implementations24 Oct 2014 Ziming Zhang, Yu-Ting Chen, Venkatesh Saligrama

We first map each pixel of an image to a visual word using a codebook, which is learned in an unsupervised manner.

Person Re-Identification

Object Proposal Generation using Two-Stage Cascade SVMs

no code implementations20 Jul 2014 Ziming Zhang, Philip H. S. Torr

Specifically, we explain our scale/aspect-ratio quantization scheme, and investigate the effects of combinations of $\ell_1$ and $\ell_2$ regularizers in cascade SVMs with/without ranking constraints in learning.

Object Proposal Generation Object Recognition +1

PRISM: Person Re-Identification via Structured Matching

no code implementations13 Jun 2014 Ziming Zhang, Venkatesh Saligrama

From a visual perspective re-id is challenging due to significant changes in visual appearance of individuals in cameras with different pose, illumination and calibration.

Graph Matching Person Re-Identification

RAPID: Rapidly Accelerated Proximal Gradient Algorithms for Convex Minimization

no code implementations13 Jun 2014 Ziming Zhang, Venkatesh Saligrama

In this paper, we propose a new algorithm to speed-up the convergence of accelerated proximal gradient (APG) methods.

BING: Binarized Normed Gradients for Objectness Estimation at 300fps

no code implementations CVPR 2014 Ming-Ming Cheng, Ziming Zhang, Wen-Yan Lin, Philip Torr

Training a generic objectness measure to produce a small set of candidate object windows, has been shown to speed up the classical sliding window object detection paradigm.

Object Detection

Regularization for Multiple Kernel Learning via Sum-Product Networks

no code implementations13 Feb 2014 Ziming Zhang

In this paper, we are interested in constructing general graph-based regularizers for multiple kernel learning (MKL) given a structure which is used to describe the way of combining basis kernels.

Learning Anchor Planes for Classification

no code implementations NeurIPS 2011 Ziming Zhang, Lubor Ladicky, Philip Torr, Amir Saffari

It provides a set of anchor points which form a local coordinate system, such that each data point on the manifold can be approximated by a linear combination of its anchor points, and the linear weights become the local coordinate coding.

Classification General Classification

Cannot find the paper you are looking for? You can Submit a new open access paper.