Models, code, and papers for "Lei Zhao":
Accurate detection of the myocardial infarction (MI) area is crucial for early diagnosis planning and follow-up management. In this study, we propose an end-to-end deep-learning algorithm framework (OF-RNN ) to accurately detect the MI area at the pixel level. Our OF-RNN consists of three different function layers: the heart localization layers, which can accurately and automatically crop the region-of-interest (ROI) sequences, including the left ventricle, using the whole cardiac magnetic resonance image sequences; the motion statistical layers, which are used to build a time-series architecture to capture two types of motion features (at the pixel-level) by integrating the local motion features generated by long short-term memory-recurrent neural networks and the global motion features generated by deep optical flows from the whole ROI sequence, which can effectively characterize myocardial physiologic function; and the fully connected discriminate layers, which use stacked auto-encoders to further learn these features, and they use a softmax classifier to build the correspondences from the motion features to the tissue identities (infarction or not) for each pixel. Through the seamless connection of each layer, our OF-RNN can obtain the area, position, and shape of the MI for each patient. Our proposed framework yielded an overall classification accuracy of 94.35% at the pixel level, from 114 clinical subjects. These results indicate the potential of our proposed method in aiding standardized MI assessments.
Nearest neighbor search and k-nearest neighbor graph construction are two fundamental issues arise from many disciplines such as information retrieval, data-mining, machine learning and computer vision. Despite continuous efforts have been taken in the last several decades, these two issues remain challenging. They become more and more imminent given the big data emerges in various fields and has been expanded significantly over the years. In this paper, a simple but effective solution both for k-nearest neighbor search and k-nearest neighbor graph construction is presented. Namely, these two issues are addressed jointly. On one hand, the k-nearest neighbor graph construction is treated as a nearest neighbor search task. Each data sample along with its k-nearest neighbors are joined into the k-nearest neighbor graph by sequentially performing the nearest neighbor search on the graph under construction. On the other hand, the built k-nearest neighbor graph is used to support k-nearest neighbor search. Since the graph is built online, dynamic updating of the graph, which is not desirable from most of the existing solutions, is supported. Moreover, this solution is feasible for various distance measures. Its effectiveness both as a k-nearest neighbor construction and k-nearest neighbor search approach is verified across various datasets in different scales, various dimensions and under different metrics.
Instance search is an interesting task as well as a challenging issue due to the lack of effective feature representation. In this paper, an instance level feature representation built upon recent fully convolutional instance-aware segmentation is proposed. The feature is ROI-pooled based on the segmented instance region. So that instances in different sizes and layouts are represented by deep feature in uniform length. This representation is further enhanced by the use of deformable ResNeXt blocks. Superior performance in terms of its distinctiveness and scalability is observed on a challenging evaluation dataset built by ourselves.
We propose a fast potential splitting Markov Chain Monte Carlo method which costs $O(1)$ time each step for sampling from equilibrium distributions (Gibbs measures) corresponding to particle systems with singular interacting kernels. We decompose the interacting potential into two parts, one is of long range but is smooth, and the other one is of short range but may be singular. To displace a particle, we first evolve a selected particle using the stochastic differential equation (SDE) under the smooth part with the idea of random batches, as commonly used in stochastic gradient Langevin dynamics. Then, we use the short range part to do a Metropolis rejection. Different from the classical Langevin dynamics, we only run the SDE dynamics with random batch for a short duration of time so that the cost in the first step is $O(p)$, where $p$ is the batch size. The cost of the rejection step is $O(1)$ since the interaction used is of short range. We justify the proposed random-batch Monte Carlo method, which combines the random batch and splitting strategies, both in theory and with numerical experiments. While giving comparable results for typical examples of the Dyson Brownian motion and Lennard-Jones fluids, our method can save more time when compared to the classical Metropolis-Hastings algorithm.
Due to the lack of suitable feature representation, effective solution to the instance search is still slow to occur. In this paper, a novel instance-level feature descriptor is proposed. The feature is built upon the salient instance region that is activated by a layer-wise back-propagation process. Such kind of region usually covers the major part an instance and represents the common patterns shared among instances of the same category. The back-propagation starts from the last convolution layer of pre-trained CNN that is originally used for classification. This makes the feature representation remain effective for instances from both known and unknown categories. Moreover, experiments show that it is effective for instance as well as image search tasks.
k-nearest neighbor graph is the fundamental data structure in many disciplines such as information retrieval, data-mining, pattern recognition and machine learning, etc. In the literature, considerable research has been focusing on how to efficiently build an approximate k-nearest neighbor graph (k-NN graph) for a fixed dataset. Unfortunately, a closely related issue to the graph construction has been long overlooked. Namely, few literature covers about how to merge two existing k-NN graphs. In this paper, we address the k-NN graph merge issue of two different scenarios. On the first hand, peer merge is proposed to address the problem of merging two approximate k-NN graphs into one. This makes parallel approximate k-NN graph computation in large-scale become possible. In addition, the problem of merging a raw set into a built k-NN graph is also addressed by joint merge. It allows the approximate k-NN graph to be built incrementally. It therefore supports approximate k-NN graph construction for an open set. Moreover, deriving from joint merge, an hierarchical approximate k-NN graph construction approach is presented. With the support of produced graph hierarchy, superior performance is observed on the large-scale NN search task across various data types and data dimensions, and under different distance measures.
Neural architecture search (NAS) is proposed to automate the architecture design process and attracts overwhelming interest from both academia and industry. However, it is confronted with overfitting issue due to the high-dimensional search space composed by $operator$ selection and $skip$ connection of each layer. This paper analyzes the overfitting issue from a novel perspective, which separates the primitives of search space into architecture-overfitting related and parameter-overfitting related elements. The $operator$ of each layer, which mainly contributes to parameter-overfitting and is important for model acceleration, is selected as our optimization target based on state-of-the-art architecture, meanwhile $skip$ which related to architecture-overfitting, is ignored. With the largely reduced search space, our proposed method is both quick to converge and practical to use in various tasks. Extensive experiments have demonstrated that the proposed method can achieve fascinated results, including classification, face recognition etc.
Hierarchical navigable small world (HNSW) graphs get more and more popular on large-scale nearest neighbor search tasks since the source codes were released two years ago. The attractiveness of this approach lies in its superior performance over most of the known nearest neighbor search approaches as well as its genericness to various distance measures. In this paper, several comparative studies have been conducted on this search approach. The role of hierarchical structure in HNSW and the function of HNSW graph itself are investigated. We find that the hierarchical structure in HNSW could not achieve "a much better logarithmic complexity scaling" as it was claimed in the paper, particularly on high dimensional data. Moreover, we find that similar high search speed efficiency as HNSW could be achieved with the support of flat k-NN graph after graph diversification. Finally, we point out the difficulty, that is faced by most of the graph based search approaches, is directly linked to "curse of dimensionality". HNSW, like other graph based approaches, is unable to address such difficulty.
Multimode fibres (MMF) are remarkable high-capacity information channels owing to the large number of transmitting fibre modes, and have recently attracted significant renewed interest in applications such as optical communication, imaging, and optical trapping. At the same time, the optical transmitting modes inside MMFs are highly sensitive to external perturbations and environmental changes, resulting in MMF transmission channels being highly variable and random. This largely limits the practical application of MMFs and hinders the full exploitation of their information capacity. Despite great research efforts made to overcome the high variability and randomness inside MMFs, any geometric change to the MMF leads to completely different transmission matrices, which unavoidably fails at the information recovery. Here, we show the successful binary image transmission using deep learning through a single MMF, which is stationary or subject to dynamic shape variations. We found that a single convolutional neural network has excellent generalisation capability with various MMF transmission states. This deep neural network can be trained by multiple MMF transmission states to accurately predict unknown information at the other end of the MMF at any of these states, without knowing which state is present. Our results demonstrate that deep learning is a promising solution to address the variability and randomness challenge of MMF based information channels. This deep-learning approach is the starting point of developing future high-capacity MMF optical systems and devices, and is applicable to optical systems concerning other diffusing media.
The great success of deep learning shows that its technology contains profound truth, and understanding its internal mechanism not only has important implications for the development of its technology and effective application in various fields, but also provides meaningful insights into the understanding of human brain mechanism. At present, most of the theoretical research on deep learning is based on mathematics. This dissertation proposes that the neural network of deep learning is a physical system, examines deep learning from three different perspectives: microscopic, macroscopic, and physical world views, answers multiple theoretical puzzles in deep learning by using physics principles. For example, from the perspective of quantum mechanics and statistical physics, this dissertation presents the calculation methods for convolution calculation, pooling, normalization, and Restricted Boltzmann Machine, as well as the selection of cost functions, explains why deep learning must be deep, what characteristics are learned in deep learning, why Convolutional Neural Networks do not have to be trained layer by layer, and the limitations of deep learning, etc., and proposes the theoretical direction and basis for the further development of deep learning now and in the future. The brilliance of physics flashes in deep learning, we try to establish the deep learning technology based on the scientific theory of physics.
Clustering analysis identifies samples as groups based on either their mutual closeness or homogeneity. In order to detect clusters in arbitrary shapes, a novel and generic solution based on boundary erosion is proposed. The clusters are assumed to be separated by relatively sparse regions. The samples are eroded sequentially according to their dynamic boundary densities. The erosion starts from low density regions, invading inwards, until all the samples are eroded out. By this manner, boundaries between different clusters become more and more apparent. It therefore offers a natural and powerful way to separate the clusters when the boundaries between them are hard to be drawn at once. With the sequential order of being eroded, the sequential boundary levels are produced, following which the clusters in arbitrary shapes are automatically reconstructed. As demonstrated across various clustering tasks, it is able to outperform most of the state-of-the-art algorithms and its performance is nearly perfect in some scenarios.
In the era of big data, k-means clustering has been widely adopted as a basic processing tool in various contexts. However, its computational cost could be prohibitively high as the data size and the cluster number are large. It is well known that the processing bottleneck of k-means lies in the operation of seeking closest centroid in each iteration. In this paper, a novel solution towards the scalability issue of k-means is presented. In the proposal, k-means is supported by an approximate k-nearest neighbors graph. In the k-means iteration, each data sample is only compared to clusters that its nearest neighbors reside. Since the number of nearest neighbors we consider is much less than k, the processing cost in this step becomes minor and irrelevant to k. The processing bottleneck is therefore overcome. The most interesting thing is that k-nearest neighbor graph is constructed by iteratively calling the fast $k$-means itself. Comparing with existing fast k-means variants, the proposed algorithm achieves hundreds to thousands times speed-up while maintaining high clustering quality. As it is tested on 10 million 512-dimensional data, it takes only 5.2 hours to produce 1 million clusters. In contrast, to fulfill the same scale of clustering, it would take 3 years for traditional k-means.
Quantization of weights of deep neural networks (DNN) has proven to be an effective solution for the purpose of implementing DNNs on edge devices such as mobiles, ASICs and FPGAs, because they have no sufficient resources to support computation involving millions of high precision weights and multiply-accumulate operations. This paper proposes a novel method to compress vectors of high precision weights of DNNs to ternary vectors, namely a cosine similarity based target non-retraining ternary (TNT) compression method. Our method leverages cosine similarity instead of Euclidean distances as commonly used in the literature and succeeds in reducing the size of the search space to find optimal ternary vectors from 3N to N, where N is the dimension of target vectors. As a result, the computational complexity for TNT to find theoretically optimal ternary vectors is only O(N log(N)). Moreover, our experiments show that, when we ternarize models of DNN with high precision parameters, the obtained quantized models can exhibit sufficiently high accuracy so that re-training models is not necessary.
Numerous methods for crafting adversarial examples were proposed recently with high success rate. Most existing works normalize images into a continuous, real vector, domain firstly, and then craft adversarial examples in this domain. However, "adversarial" examples may become benign after de-normalizing them back into the discrete integer domain, known as the discretization problem. The discretization problem was mentioned in some work, but was underestimated and has received relatively little attention. In this work, we conduct the first comprehensive study of the discretization problem. We theoretically analyze 34 representative methods and empirically study 20 representative open source tools for crafting adversarial images. Our study reveals that almost all existing works suffer from the discretization problem and it is far more serious than originally thought. For instance, most black-box methods downgrade to white-box ones and methods having higher success rates drop down to lower high success rates, e.g., from 100% to 10%. This suggests that the discretization problem should be taken into account when crafting adversarial examples. As a first step towards addressing this problem, we propose a black-box method which reduces the adversarial example searching problem to a derivative-free optimization problem. Our method is able to craft `real' adversarial images by derivative-free search on the discrete integer domain. Experimental results show that our method achieves significantly higher success rate in terms of adversarial examples in the discrete integer domain than most other methods, no matter white-box or black-box. Moreover, our method is able to handle models that is non-differentiable and we successfully break the winner of NIPS 17 competition on defense with 95% success rate.
Recent studies using deep neural networks have shown remarkable success in style transfer especially for artistic and photo-realistic images. However, the approaches using global feature correlations fail to capture small, intricate textures and maintain correct texture scales of the artworks, and the approaches based on local patches are defective on global effect. In this paper, we present a novel feature pyramid fusion neural network, dubbed GLStyleNet, which sufficiently takes into consideration multi-scale and multi-level pyramid features by best aggregating layers across a VGG network, and performs style transfer hierarchically with multiple losses of different scales. Our proposed method retains high-frequency pixel information and low frequency construct information of images from two aspects: loss function constraint and feature fusion. Our approach is not only flexible to adjust the trade-off between content and style, but also controllable between global and local. Compared to state-of-the-art methods, our method can transfer not just large-scale, obvious style cues but also subtle, exquisite ones, and dramatically improves the quality of style transfer. We demonstrate the effectiveness of our approach on portrait style transfer, artistic style transfer, photo-realistic style transfer and Chinese ancient painting style transfer tasks. Experimental results indicate that our unified approach improves image style transfer quality over previous state-of-the-art methods, while also accelerating the whole process in a certain extent. Our code is available at https://github.com/EndyWon/GLStyleNet.
Nearest neighbor search is known as a challenging issue that has been studied for several decades. Recently, this issue becomes more and more imminent in viewing that the big data problem arises from various fields. In this paper, a scalable solution based on hill-climbing strategy with the support of k-nearest neighbor graph (kNN) is presented. Two major issues have been considered in the paper. Firstly, an efficient kNN graph construction method based on two means tree is presented. For the nearest neighbor search, an enhanced hill-climbing procedure is proposed, which sees considerable performance boost over original procedure. Furthermore, with the support of inverted indexing derived from residue vector quantization, our method achieves close to 100% recall with high speed efficiency in two state-of-the-art evaluation benchmarks. In addition, a comparative study on both the compressional and traditional nearest neighbor search methods is presented. We show that our method achieves the best trade-off between search quality, efficiency and memory complexity.
Inspired by the recent advances in deep learning (DL), this work presents a deep neural network aided decoding algorithm for binary linear codes. Based on the concept of deep unfolding, we design a decoding network by unfolding the alternating direction method of multipliers (ADMM)-penalized decoder. In addition, we propose two improved versions of the proposed network. The first one transforms the penalty parameter into a set of iteration-dependent ones, and the second one adopts a specially designed penalty function, which is based on a piecewise linear function with adjustable slopes. Numerical results show that the resulting DL-aided decoders outperform the original ADMM-penalized decoder for various low density parity check (LDPC) codes with similar computational complexity.
Existing domain adaptation methods on visual sentiment classification typically are investigated under the single-source scenario, where the knowledge learned from a source domain of sufficient labeled data is transferred to the target domain of loosely labeled or unlabeled data. However, in practice, data from a single source domain usually have a limited volume and can hardly cover the characteristics of the target domain. In this paper, we propose a novel multi-source domain adaptation (MDA) method, termed Multi-source Sentiment Generative Adversarial Network (MSGAN), for visual sentiment classification. To handle data from multiple source domains, it learns to find a unified sentiment latent space where data from both the source and target domains share a similar distribution. This is achieved via cycle consistent adversarial learning in an end-to-end manner. Extensive experiments conducted on four benchmark datasets demonstrate that MSGAN significantly outperforms the state-of-the-art MDA approaches for visual sentiment classification.
In this work, we consider the use of model-driven deep learning techniques for massive multiple-input multiple-output (MIMO) detection. Compared with conventional MIMO systems, massive MIMO promises improved spectral efficiency, coverage and range. Unfortunately, these benefits are coming at the cost of significantly increased computational complexity. To reduce the complexity of signal detection and guarantee the performance, we present a learned conjugate gradient descent network (LcgNet), which is constructed by unfolding the iterative conjugate gradient descent (CG) detector. In the proposed network, instead of calculating the exact values of the scalar step-sizes, we explicitly learn their universal values. Also, we can enhance the proposed network by augmenting the dimensions of these step-sizes. Furthermore, in order to reduce the memory costs, a novel quantized LcgNet is proposed, where a low-resolution nonuniform quantizer is integrated into the LcgNet to smartly quantize the aforementioned step-sizes. The quantizer is based on a specially designed soft staircase function with learnable parameters to adjust its shape. Meanwhile, due to fact that the number of learnable parameters is limited, the proposed networks are easy and fast to train. Numerical results demonstrate that the proposed network can achieve promising performance with much lower complexity.