We consider image transformation problems, where an input image is transformed into an output image. Recent methods for such problems typically train feed-forward convolutional neural networks using a per-pixel loss between the output and ground-truth images. Parallel work has shown that high-quality images can be generated by defining and optimizing perceptual loss functions based on high-level features extracted from pretrained networks. We combine the benefits of both approaches, and propose the use of perceptual loss functions for training feed-forward networks for image transformation tasks. We show results on image style transfer, where a feed-forward network is trained to solve the optimization problem proposed by Gatys et al. in real-time. Compared to the optimization-based method, our network gives similar qualitative results but is three orders of magnitude faster. We also experiment with single-image super-resolution, where replacing a per-pixel loss with a perceptual loss gives visually pleasing results.

/pdf/perceptual-losses-for-real-time-style-transfer-and-super-hq4q54uf80.pdf

Perceptual Losses for Real-Time Style Transfer and Super-Resolution

We propose a deep learning method for single image super-resolution (SR). Our method directly learns an end-to-end mapping between the low/high-resolution images. The mapping is represented as a deep convolutional neural network (CNN) that takes the low-resolution image as the input and outputs the high-resolution one. We further show that traditional sparse-coding-based SR methods can also be viewed as a deep convolutional network. But unlike traditional methods that handle each component separately, our method jointly optimizes all layers. Our deep CNN has a lightweight structure, yet demonstrates state-of-the-art restoration quality, and achieves fast speed for practical on-line usage. We explore different network structures and parameter settings to achieve trade-offs between performance and speed. Moreover, we extend our network to cope with three color channels simultaneously, and show better overall reconstruction quality.

Image Super-Resolution Using Deep Convolutional Networks

Data analysis plays an indispensable role for understanding various phenomena. Cluster analysis, primitive exploration with little or no prior knowledge, consists of research developed across a wide variety of communities. The diversity, on one hand, equips us with many tools. On the other hand, the profusion of options causes confusion. We survey clustering algorithms for data sets appearing in statistics, computer science, and machine learning, and illustrate their applications in some benchmark data sets, the traveling salesman problem, and bioinformatics, a new field attracting intensive efforts. Several tightly related topics, proximity measure, and cluster validation, are also discussed.

/pdf/survey-of-clustering-algorithms-6zezqhbjyo.pdf

Survey of clustering algorithms

We consider image transformation problems, where an input image is transformed into an output image. Recent methods for such problems typically train feed-forward convolutional neural networks using a \emph{per-pixel} loss between the output and ground-truth images. Parallel work has shown that high-quality images can be generated by defining and optimizing \emph{perceptual} loss functions based on high-level features extracted from pretrained networks. We combine the benefits of both approaches, and propose the use of perceptual loss functions for training feed-forward networks for image transformation tasks. We show results on image style transfer, where a feed-forward network is trained to solve the optimization problem proposed by Gatys et al in real-time. Compared to the optimization-based method, our network gives similar qualitative results but is three orders of magnitude faster. We also experiment with single-image super-resolution, where replacing a per-pixel loss with a perceptual loss gives visually pleasing results.

This paper presents a new approach to single-image superresolution, based upon sparse signal representation. Research on image statistics suggests that image patches can be well-represented as a sparse linear combination of elements from an appropriately chosen over-complete dictionary. Inspired by this observation, we seek a sparse representation for each patch of the low-resolution input, and then use the coefficients of this representation to generate the high-resolution output. Theoretical results from compressed sensing suggest that under mild conditions, the sparse representation can be correctly recovered from the downsampled signals. By jointly training two dictionaries for the low- and high-resolution image patches, we can enforce the similarity of sparse representations between the low-resolution and high-resolution image patch pair with respect to their own dictionaries. Therefore, the sparse representation of a low-resolution image patch can be applied with the high-resolution image patch dictionary to generate a high-resolution image patch. The learned dictionary pair is a more compact representation of the patch pairs, compared to previous approaches, which simply sample a large amount of image patch pairs , reducing the computational cost substantially. The effectiveness of such a sparsity prior is demonstrated for both general image super-resolution (SR) and the special case of face hallucination. In both cases, our algorithm generates high-resolution images that are competitive or even superior in quality to images produced by other similar SR methods. In addition, the local sparse modeling of our approach is naturally robust to noise, and therefore the proposed algorithm can handle SR with noisy inputs in a more unified framework.

/pdf/image-super-resolution-via-sparse-representation-18yd1f09t4.pdf

Image Super-Resolution Via Sparse Representation

In this paper, we propose a novel method for solving single-image super-resolution problems. Given a low-resolution image as input, we recover its high-resolution counterpart using a set of training examples. While this formulation resembles other learning-based methods for super-resolution, our method has been inspired by recent manifold teaming methods, particularly locally linear embedding (LLE). Specifically, small image patches in the lowand high-resolution images form manifolds with similar local geometry in two distinct feature spaces. As in LLE, local geometry is characterized by how a feature vector corresponding to a patch can be reconstructed by its neighbors in the feature space. Besides using the training image pairs to estimate the high-resolution embedding, we also enforce local compatibility and smoothness constraints between patches in the target high-resolution image through overlapping. Experiments show that our method is very flexible and gives good empirical results.

/pdf/super-resolution-through-neighbor-embedding-3fq6pbub3u.pdf

Super-resolution through neighbor embedding

sssHandwritten signature is the most widely accepted biometric for identity verification. To facilitate objective evaluation and comparison of algorithms in the field of automatic handwritten signature verification, we organized the First International Signature Verification Competition (SVC2004) recently as a step towards establishing common benchmark databases and benchmarking rules. For each of the two tasks of the competition, a signature database involving 100 sets of signature data was created, with 20 genuine signatures and 20 skilled forgeries for each set. Eventually, 13 teams competed for Task 1 and eight teams competed for Task 2. When evaluated on data with skilled forgeries, the best team for Task 1 gives an equal error rate (EER) of 2.84% and that for Task 2 gives an EER of 2.89%. We believe that SVC2004 has successfully achieved its goals and the experience gained from SVC2004 will be very useful to similar activities in the future.

/pdf/svc2004-first-international-signature-verification-qry541070q.pdf

SVC2004: First International Signature Verification Competition

https://www.cse.ust.hk/faculty/dyyeung/paper/pdf/yeung.pr2004.pdf

Time Series Clustering with ARMA Mixtures

Clustering problems are central to many knowledge discovery and data mining tasks. However, most existing clustering methods can only work with fixed-dimensional representations of data patterns. In this paper we study the clustering of data patterns that are represented as sequences or time series possibly of different lengths. We propose a model-based approach to this problem using mixtures of autoregressive moving average (ARMA) models. We derive an expectation-maximization (EM) algorithm for learning the mixing coefficients as well as the parameters of component models. Experiments were conducted on simulated and real datasets. Results show that our method compares favorably with another method recently proposed by others for similar time series clustering problems.

/pdf/mixtures-of-arma-models-for-model-based-time-series-3oqeu5eu28.pdf

Mixtures of ARMA models for model-based time series clustering

Extremile regression proposed in recent years not only retains the advantage of quantile regression that can fully show the information of sample data by setting different quantiles, but also has its own superiority compared with quantile regression and expectile regression, due to its explicit expression and conservativeness in estimating. In this paper, we propose a linear extremile regression model and introduce a variable selection method using a penalty called quasi elastic net (QEN) to solve high-dimensional problems. Moreover, we propose an EM algorithm and establish corresponding theoretical properties under some mild conditions. In numerical studies, we compare QEN penalty with <inline-formula><tex-math id="M1">\begin{document}$L_{0}$\end{document}</tex-math><alternatives><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="JUSTC-2022-0099_M1.jpg"/><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="JUSTC-2022-0099_M1.png"/></alternatives></inline-formula>, <inline-formula><tex-math id="M2">\begin{document}$L_{1}$\end{document}</tex-math><alternatives><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="JUSTC-2022-0099_M2.jpg"/><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="JUSTC-2022-0099_M2.png"/></alternatives></inline-formula>, <inline-formula><tex-math id="M3">\begin{document}$L_{2}$\end{document}</tex-math><alternatives><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="JUSTC-2022-0099_M3.jpg"/><graphic xmlns:xlink="http://www.w3.org/1999/xlink" xlink:href="JUSTC-2022-0099_M3.png"/></alternatives></inline-formula> and elastic net penalty and the results show that the proposed method is effective and has certain advantages in analysis.

Yimin Xiong

Papers

Super-resolution through neighbor embedding

SVC2004: First International Signature Verification Competition

Time Series Clustering with ARMA Mixtures

Mixtures of ARMA models for model-based time series clustering

Variable selection in high-dimensional extremile regression via the quasi elastic net