arXiv subjects

Shaohua Li

Publications and source records attributed to Shaohua Li.

At least 55 records · Page 3Linked to original sources

arXiv2020

Many visits TSP revisited

We study the Many Visits TSP problem, where given a number $k(v)$ for each of $n$ cities and pairwise (possibly asymmetric) integer distances, one has to find an optimal tour that visits each city $v$ exactly $k(v)$ times. The currently fastest algorithm is due to Berger, Kozma, Mnich and Vincze [SODA 2019, TALG 2020] and runs in time and space $\mathcal{O}^*(5^n)$. They also show a polynomial space algorithm running in time $\mathcal{O}^*(16^{n+o(n)})$. In this work, we show three main results: (i) A randomized polynomial space algorithm in time $\mathcal{O}^*(2^nD)$, where $D$ is the maximum distance between two cities. By using standard methods, this results in $(1+ε)$-approximation in time $\mathcal{O}^*(2^nε^{-1})$. Improving the constant $2$ in these results would be a major breakthrough, as it would result in improving the $\mathcal{O}^*(2^n)$-time algorithm for Directed Hamiltonian Cycle, which is a 50 years old open problem. (ii) A tight analysis of Berger et al.'s exponential space algorithm, resulting in $\mathcal{O}^*(4^n)$ running time bound. (iii) A new polynomial space algorithm, running in time $\mathcal{O}(7.88^n)$.

cs.DS↗

arXiv2020

Feature Lenses: Plug-and-play Neural Modules for Transformation-Invariant Visual Representations

Convolutional Neural Networks (CNNs) are known to be brittle under various image transformations, including rotations, scalings, and changes of lighting conditions. We observe that the features of a transformed image are drastically different from the ones of the original image. To make CNNs more invariant to transformations, we propose "Feature Lenses", a set of ad-hoc modules that can be easily plugged into a trained model (referred to as the "host model"). Each individual lens reconstructs the original features given the features of a transformed image under a particular transformation. These lenses jointly counteract feature distortions caused by various transformations, thus making the host model more robust without retraining. By only updating lenses, the host model is freed from iterative updating when facing new transformations absent in the training data; as feature semantics are preserved, downstream applications, such as classifiers and detectors, automatically gain robustness without retraining. Lenses are trained in a self-supervised fashion with no annotations, by minimizing a novel "Top-K Activation Contrast Loss" between lens-transformed features and original features. Evaluated on ImageNet, MNIST-rot, and CIFAR-10, Feature Lenses show clear advantages over baseline methods.

cs.CV↗

arXiv2020

An improved FPT algorithm for Independent Feedback Vertex Set

We study the Independent Feedback Vertex Set problem - a variant of the classic Feedback Vertex Set problem where, given a graph $G$ and an integer $k$, the problem is to decide whether there exists a vertex set $S\subseteq V(G)$ such that $G\setminus S$ is a forest and $S$ is an independent set of size at most $k$. We present an $O^\ast((1+φ^{2})^{k})$-time FPT algorithm for this problem, where $φ<1.619$ is the golden ratio, improving the previous fastest $O^\ast(4.1481^{k})$-time algorithm given by Agrawal et al [IPEC 2016]. The exponential factor in our time complexity bound matches the fastest deterministic FPT algorithm for the classic Feedback Vertex Set problem. On the technical side, the main novelty is a refined measure of an input instance in a branching process, that allows for a simpler and more concise description and analysis of the algorithm.

cs.DS↗

arXiv2019

Illumination Robust Loop Closure Detection with the Constraint of Pose

Background: Loop closure detection is a crucial part in robot navigation and simultaneous location and mapping (SLAM). Appearance-based loop closure detection still faces many challenges, such as illumination changes, perceptual aliasing and increasing computational complexity. Method: In this paper, we proposed a visual loop-closure detection algorithm which combines illumination robust descriptor DIRD and odometry information. The estimated pose and variance are calculated by the visual inertial odometry (VIO), then the loop closure candidate areas are found based on the distance between images. We use a new distance combing the the Euclidean distance and the Mahalanobis distance and a dynamic threshold to select the loop closure candidate areas. Finally, in loop-closure candidate areas, we do image retrieval with DIRD which is an illumination robust descriptor. Results: The proposed algorithm is evaluated on KITTI_00 and EuRoc datasets. The results show that the loop closure areas could be correctly detected and the time consumption is effectively reduced. We compare it with SeqSLAM algorithm, the proposed algorithm gets better performance on PR-curve.

cs.RO↗

arXiv2019

Invariant Cubature Kalman Filter for Monocular Visual Inertial Odometry with Line Features

To achieve robust and accurate state estimation for robot navigation, we propose a novel Visual Inertial Odometry(VIO) algorithm with line features upon the theory of invariant Kalman filtering and Cubature Kalman Filter (CKF). In contrast with conventional CKF, the state of the filter is constructed by a high dimensional Matrix Lie group and the uncertainty is represented using Lie algebra. To improve the robustness of system in challenging scenes, e.g. low-texture or illumination changing environments, line features are brought into the state variable. In the proposed algorithm, exponential mapping of Lie algebra is used to construct the cubature points and the re-projection errors of lines are built as observation function for updating the state. This method accurately describes the system uncertainty in rotation and reduces the linearization error of system, which extends traditional CKF from Euclidean space to manifold. It not only inherits the advantages of invariant filtering in consistency, but also avoids the complex Jacobian calculation of high-dimensional matrix. To demonstrate the effectiveness of the proposed algorithm, we compare it with the state-of-the-art filtering-based VIO algorithms on Euroc datasets. And the results show that the proposed algorithm is effective in improving accuracy and robustness of estimation.

cs.RO↗

arXiv2019

RoboCoDraw: Robotic Avatar Drawing with GAN-based Style Transfer and Time-efficient Path Optimization

Robotic drawing has become increasingly popular as an entertainment and interactive tool. In this paper we present RoboCoDraw, a real-time collaborative robot-based drawing system that draws stylized human face sketches interactively in front of human users, by using the Generative Adversarial Network (GAN)-based style transfer and a Random-Key Genetic Algorithm (RKGA)-based path optimization. The proposed RoboCoDraw system takes a real human face image as input, converts it to a stylized avatar, then draws it with a robotic arm. A core component in this system is the Avatar-GAN proposed by us, which generates a cartoon avatar face image from a real human face. AvatarGAN is trained with unpaired face and avatar images only and can generate avatar images of much better likeness with human face images in comparison with the vanilla CycleGAN. After the avatar image is generated, it is fed to a line extraction algorithm and converted to sketches. An RKGA-based path optimization algorithm is applied to find a time-efficient robotic drawing path to be executed by the robotic arm. We demonstrate the capability of RoboCoDraw on various face images using a lightweight, safe collaborative robot UR5.

cs.RO↗

arXiv2019

Learning to Reconstruct Crack Profiles for Eddy Current Nondestructive Testing

Eddy current testing (ECT) is one of the most popular Nondestructive Testing (NDT) techniques, especially for conductive materials. Reconstructing the crack profile from measured EC signals is one of the main goals of ECT. This task is highly challenging, as the EC signals are nonlinear responses resulted from the presence of cracks, and reconstructing the crack profile requires establishing the forward model of the nonlinear electromagnetic dynamics and solving its inverse problem, which is an ill-posed numerical optimization problem. Instead of solving the inverse problem numerically, we propose to directly learn the inverse mapping from EC signals to crack profiles with a deep encoder-decoder convolutional neural network named EddyNet. EddyNet is trained on a set of randomly generated crack profiles and the corresponding simulated EC responses generated from a realistic forward model. On the held-out test data, EddyNet achieved a mean absolute error of 0.198 between predicted profiles and ground truth ones. Qualitatively, the geometries of predicted profiles are visually similar to the ground truth profiles. Our method greatly reduces the usual reliance on domain experts, and the reconstruction is extremely fast both on GPUs and on CPUs. The source code of EddyNet is released on https://github.com/askerlee/EddyNet.

eess.IV↗

arXiv2019

Multi-Instance Multi-Scale CNN for Medical Image Classification

Deep learning for medical image classification faces three major challenges: 1) the number of annotated medical images for training are usually small; 2) regions of interest (ROIs) are relatively small with unclear boundaries in the whole medical images, and may appear in arbitrary positions across the x,y (and also z in 3D images) dimensions. However often only labels of the whole images are annotated, and localized ROIs are unavailable; and 3) ROIs in medical images often appear in varying sizes (scales). We approach these three challenges with a Multi-Instance Multi-Scale (MIMS) CNN: 1) We propose a multi-scale convolutional layer, which extracts patterns of different receptive fields with a shared set of convolutional kernels, so that scale-invariant patterns are captured by this compact set of kernels. As this layer contains only a small number of parameters, training on small datasets becomes feasible; 2) We propose a "top-k pooling" to aggregate the feature maps in varying scales from multiple spatial dimensions, allowing the model to be trained using weak annotations within the multiple instance learning (MIL) framework. Our method is shown to perform well on three classification tasks involving two 3D and two 2D medical image datasets.

cs.CV↗

arXiv2019

An Improved FPT Algorithm for the Flip Distance Problem

Given a set $\cal P$ of points in the Euclidean plane and two triangulations of $\cal P$, the flip distance between these two triangulations is the minimum number of flips required to transform one triangulation into the other. Parameterized Flip Distance problem is to decide if the flip distance between two given triangulations is equal to a given integer $k$. The previous best FPT algorithm runs in time $O^{*}(k\cdot c^{k})$ ($c\leq 2\times 14^{11}$), where each step has fourteen possible choices, and the length of the action sequence is bounded by $11k$. By applying the backtracking strategy and analyzing the underlying property of the flip sequence, each step of our algorithm has only five possible choices. Based on an auxiliary graph $G$, we prove that the length of the action sequence for our algorithm is bounded by $2|G|$. As a result, we present an FPT algorithm running in time $O^{*}(k\cdot 32^{k})$.

cs.DS↗

arXiv2019

Feature Aggregation Network for Video Face Recognition

This paper aims to learn a compact representation of a video for video face recognition task. We make the following contributions: first, we propose a meta attention-based aggregation scheme which adaptively and fine-grained weighs the feature along each feature dimension among all frames to form a compact and discriminative representation. It makes the best to exploit the valuable or discriminative part of each frame to promote the performance of face recognition, without discarding or despising low quality frames as usual methods do. Second, we build a feature aggregation network comprised of a feature embedding module and a feature aggregation module. The embedding module is a convolutional neural network used to extract a feature vector from a face image, while the aggregation module consists of cascaded two meta attention blocks which adaptively aggregate the feature vectors into a single fixed-length representation. The network can deal with arbitrary number of frames, and is insensitive to frame order. Third, we validate the performance of proposed aggregation scheme. Experiments on publicly available datasets, such as YouTube face dataset and IJB-A dataset, show the effectiveness of our method, and it achieves competitive performances on both the verification and identification protocols.

cs.CV↗

arXiv2019

Multi-budgeted directed cuts

We study multi-budgeted variants of the classic minimum cut problem and graph separation problems that turned out to be important in parameterized complexity: Skew Multicut and Directed Feedback Arc Set. In our generalization, we assign colors $1,2,...,\ell$ to some edges and give separate budgets $k_{1},k_{2},...,k_{\ell}$. Let $E_{i}$ be the set of edges of color $i$. The solution $C$ for the multi-budgeted variant of a graph separation problem not only needs to satisfy the usual separation requirements, but also needs to satisfy that $|C\cap E_{i}|\leq k_{i}$ for every $i\in \{1,...,\ell\}$. Contrary to the classic minimum cut problem, the multi-budgeted variant turns out to be NP-hard even for $\ell = 2$. We propose FPT algorithms parameterized by $k=k_{1}+...+k_{\ell}$ for all three problems. To this end, we develop a branching procedure for the multi-budgeted minimum cut problem that measures the progress of the algorithm not by reducing $k$ as usual, by but elevating the capacity of some edges and thus increasing the size of maximum source-to-sink flow. Using the fact that a similar strategy is used to enumerate all important separators of a given size, we merge this process with the flow-guided branching and show an FPT bound on the number of (appropriately defined) important multi-budgeted separators. This allows us to extend our algorithm to the Skew Multicut and Directed Feedback Arc Set problems. Furthermore, we show connections of the multi-budgeted variants with weighted variants of the directed cut problems and the Chain $\ell$-SAT problem, whose parameterized complexity remains an open problem. We show that these problems admit a bounded-in-parameter number of "maximally pushed" solutions (in a similar spirit as important separators are maximally pushed), giving somewhat weak evidence towards their tractability.

cs.DS↗

arXiv2018

FALCON: A Fourier Transform Based Approach for Fast and Secure Convolutional Neural Network Predictions

Machine learning as a service has been widely deployed to utilize deep neural network models to provide prediction services. However, this raises privacy concerns since clients need to send sensitive information to servers. In this paper, we focus on the scenario where clients want to classify private images with a convolutional neural network model hosted in the server, while both parties keep their data private. We present FALCON, a fast and secure approach for CNN predictions based on Fourier Transform. Our solution enables linear layers of a CNN model to be evaluated simply and efficiently with fully homomorphic encryption. We also introduce the first efficient and privacy-preserving protocol for softmax function, which is an indispensable component in CNNs and has not yet been evaluated in previous works due to its high complexity. We implemented the FALCON and evaluated the performance on real-world CNN models. The experimental results show that FALCON outperforms the best known works in both computation and communication cost.

cs.CR↗

arXiv2018

SecGrid: A Secure and Efficient SGX-enabled Smart Grid System with Rich Functionalities

Smart grid adopts two-way communication and rich functionalities to gain a positive impact on the sustainability and efficiency of power usage, but on the other hand, also poses serious challenges to customers' privacy. Existing solutions in smart grid usually use cryptographic tools, such as homomorphic encryption, to protect individual privacy, which, however, can only support limited and simple functionalities. Moreover, the resource-constrained smart meters need to perform heavy asymmetric cryptography in these solutions, which is not applied to smart grid. In this paper, we present a practical and secure SGX-enabled smart grid system, named SecGrid. Our system leverage trusted hardware SGX to ensure that grid utilities can efficiently execute rich functionalities on customers' private data, while guaranteeing their privacy. With the designed security protocols, the SecGrid only require the smart meters to perform AES encryption. Security analysis shows that SecGrid can thwart various attacks from malicious adversaries. Experimental results show that SecGrid is much faster than the existing privacy-preserving schemes in smart grid.

cs.CR↗

arXiv2017

Enhanced Attacks on Defensively Distilled Deep Neural Networks

Deep neural networks (DNNs) have achieved tremendous success in many tasks of machine learning, such as the image classification. Unfortunately, researchers have shown that DNNs are easily attacked by adversarial examples, slightly perturbed images which can mislead DNNs to give incorrect classification results. Such attack has seriously hampered the deployment of DNN systems in areas where security or safety requirements are strict, such as autonomous cars, face recognition, malware detection. Defensive distillation is a mechanism aimed at training a robust DNN which significantly reduces the effectiveness of adversarial examples generation. However, the state-of-the-art attack can be successful on distilled networks with 100% probability. But it is a white-box attack which needs to know the inner information of DNN. Whereas, the black-box scenario is more general. In this paper, we first propose the epsilon-neighborhood attack, which can fool the defensively distilled networks with 100% success rate in the white-box setting, and it is fast to generate adversarial examples with good visual quality. On the basis of this attack, we further propose the region-based attack against defensively distilled DNNs in the black-box setting. And we also perform the bypass attack to indirectly break the distillation defense as a complementary method. The experimental results show that our black-box attacks have a considerable success rate on defensively distilled networks.

cs.CV↗

arXiv2017

Laplacian-Steered Neural Style Transfer

Neural Style Transfer based on Convolutional Neural Networks (CNN) aims to synthesize a new image that retains the high-level structure of a content image, rendered in the low-level texture of a style image. This is achieved by constraining the new image to have high-level CNN features similar to the content image, and lower-level CNN features similar to the style image. However in the traditional optimization objective, low-level features of the content image are absent, and the low-level features of the style image dominate the low-level detail structures of the new image. Hence in the synthesized image, many details of the content image are lost, and a lot of inconsistent and unpleasing artifacts appear. As a remedy, we propose to steer image synthesis with a novel loss function: the Laplacian loss. The Laplacian matrix ("Laplacian" in short), produced by a Laplacian operator, is widely used in computer vision to detect edges and contours. The Laplacian loss measures the difference of the Laplacians, and correspondingly the difference of the detail structures, between the content image and a new image. It is flexible and compatible with the traditional style transfer constraints. By incorporating the Laplacian loss, we obtain a new optimization objective for neural style transfer named Lapstyle. Minimizing this objective will produce a stylized image that better preserves the detail structures of the content image and eliminates the artifacts. Experiments show that Lapstyle produces more appealing stylized images with less artifacts, without compromising their "stylishness".

cs.CV↗

arXiv2017

Dirichlet-vMF Mixture Model

This document is about the multi-document Von-Mises-Fisher mixture model with a Dirichlet prior, referred to as VMFMix. VMFMix is analogous to Latent Dirichlet Allocation (LDA) in that they can capture the co-occurrence patterns acorss multiple documents. The difference is that in VMFMix, the topic-word distribution is defined on a continuous n-dimensional hypersphere. Hence VMFMix is used to derive topic embeddings, i.e., representative vectors, from multiple sets of embedding vectors. An efficient Variational Expectation-Maximization inference algorithm is derived. The performance of VMFMix on two document classification tasks is reported, with some preliminary analysis.

cs.CL↗

arXiv2017

Document Visualization using Topic Clouds

Traditionally a document is visualized by a word cloud. Recently, distributed representation methods for documents have been developed, which map a document to a set of topic embeddings. Visualizing such a representation is useful to present the semantics of a document in higher granularity; it is also challenging, as there are multiple topics, each containing multiple words. We propose to visualize a set of topics using Topic Cloud, which is a pie chart consisting of topic slices, where each slice contains important words in this topic. To make important topics/words visually prominent, the sizes of topic slices and word fonts are proportional to their importance in the document. A topic cloud can help the user quickly evaluate the quality of derived document representations. For NLP practitioners, It can be used to qualitatively compare the topic quality of different document representation algorithms, or to inspect how model parameters impact the derived representations.

cs.IR↗

arXiv2016

Generative Topic Embedding: a Continuous Representation of Documents (Extended Version with Proofs)

Word embedding maps words into a low-dimensional continuous embedding space by exploiting the local word collocation patterns in a small context window. On the other hand, topic modeling maps documents onto a low-dimensional topic space, by utilizing the global word collocation patterns in the same document. These two types of patterns are complementary. In this paper, we propose a generative topic embedding model to combine the two types of patterns. In our model, topics are represented by embedding vectors, and are shared across documents. The probability of each word is influenced by both its local context and its topic. A variational inference method yields the topic embeddings as well as the topic mixing proportions for each document. Jointly they represent the document in a low-dimensional continuous space. In two document classification tasks, our method performs better than eight existing methods, with fewer features. In addition, we illustrate with an example that our method can generate coherent topics even based on only one document.

cs.CL↗