July 28, 2019

3126 words 15 mins read

Paper Group ANR 246

On Noisy Negative Curvature Descent: Competing with Gradient Descent for Faster Non-convex Optimization. A general unified framework for interval pairwise comparison matrices. Virtual PET Images from CT Data Using Deep Convolutional Networks: Initial Results. Deriving Compact Laws Based on Algebraic Formulation of a Data Set. Multi-Task Learning Us …

On Noisy Negative Curvature Descent: Competing with Gradient Descent for Faster Non-convex Optimization


Title	On Noisy Negative Curvature Descent: Competing with Gradient Descent for Faster Non-convex Optimization
Authors	Mingrui Liu, Tianbao Yang
Abstract	The Hessian-vector product has been utilized to find a second-order stationary solution with strong complexity guarantee (e.g., almost linear time complexity in the problem’s dimensionality). In this paper, we propose to further reduce the number of Hessian-vector products for faster non-convex optimization. Previous algorithms need to approximate the smallest eigen-value with a sufficient precision (e.g., $\epsilon_2\ll 1$) in order to achieve a sufficiently accurate second-order stationary solution (i.e., $\lambda_{\min}(\nabla^2 f(\x))\geq -\epsilon_2)$. In contrast, the proposed algorithms only need to compute the smallest eigen-vector approximating the corresponding eigen-value up to a small power of current gradient’s norm. As a result, it can dramatically reduce the number of Hessian-vector products during the course of optimization before reaching first-order stationary points (e.g., saddle points). The key building block of the proposed algorithms is a novel updating step named the NCG step, which lets a noisy negative curvature descent compete with the gradient descent. We show that the worst-case time complexity of the proposed algorithms with their favorable prescribed accuracy requirements can match the best in literature for achieving a second-order stationary point but with an arguably smaller per-iteration cost. We also show that the proposed algorithms can benefit from inexact Hessian by developing their variants accepting inexact Hessian under a mild condition for achieving the same goal. Moreover, we develop a stochastic algorithm for a finite or infinite sum non-convex optimization problem. To the best of our knowledge, the proposed stochastic algorithm is the first one that converges to a second-order stationary point in {\it high probability} with a time complexity independent of the sample size and almost linear in dimensionality.
Tasks
Published	2017-09-25
URL	http://arxiv.org/abs/1709.08571v2
PDF	http://arxiv.org/pdf/1709.08571v2.pdf
PWC	https://paperswithcode.com/paper/on-noisy-negative-curvature-descent-competing
Repo
Framework

A general unified framework for interval pairwise comparison matrices


Title	A general unified framework for interval pairwise comparison matrices
Authors	Bice Cavallo, Matteo Brunelli
Abstract	Interval Pairwise Comparison Matrices have been widely used to account for uncertain statements concerning the preferences of decision makers. Several approaches have been proposed in the literature, such as multiplicative and fuzzy interval matrices. In this paper, we propose a general unified approach to Interval Pairwise Comparison Matrices, based on Abelian linearly ordered groups. In this framework, we generalize some consistency conditions provided for multiplicative and/or fuzzy interval pairwise comparison matrices and provide inclusion relations between them. Then, we provide a concept of distance between intervals that, together with a notion of mean defined over real continuous Abelian linearly ordered groups, allows us to provide a consistency index and an indeterminacy index. In this way, by means of suitable isomorphisms between Abelian linearly ordered groups, we will be able to compare the inconsistency and the indeterminacy of different kinds of Interval Pairwise Comparison Matrices, e.g. multiplicative, additive, and fuzzy, on a unique Cartesian coordinate system.
Tasks
Published	2017-11-26
URL	http://arxiv.org/abs/1711.09441v1
PDF	http://arxiv.org/pdf/1711.09441v1.pdf
PWC	https://paperswithcode.com/paper/a-general-unified-framework-for-interval
Repo
Framework

Virtual PET Images from CT Data Using Deep Convolutional Networks: Initial Results


Title	Virtual PET Images from CT Data Using Deep Convolutional Networks: Initial Results
Authors	Avi Ben-Cohen, Eyal Klang, Stephen P. Raskin, Michal Marianne Amitai, Hayit Greenspan
Abstract	In this work we present a novel system for PET estimation using CT scans. We explore the use of fully convolutional networks (FCN) and conditional generative adversarial networks (GAN) to export PET data from CT data. Our dataset includes 25 pairs of PET and CT scans where 17 were used for training and 8 for testing. The system was tested for detection of malignant tumors in the liver region. Initial results look promising showing high detection performance with a TPR of 92.3% and FPR of 0.25 per case. Future work entails expansion of the current system to the entire body using a much larger dataset. Such a system can be used for tumor detection and drug treatment evaluation in a CT-only environment instead of the expansive and radioactive PET-CT scan.
Tasks
Published	2017-07-30
URL	http://arxiv.org/abs/1707.09585v1
PDF	http://arxiv.org/pdf/1707.09585v1.pdf
PWC	https://paperswithcode.com/paper/virtual-pet-images-from-ct-data-using-deep
Repo
Framework

Deriving Compact Laws Based on Algebraic Formulation of a Data Set


Title	Deriving Compact Laws Based on Algebraic Formulation of a Data Set
Authors	Wenqing Xu, Mark Stalzer
Abstract	In various subjects, there exist compact and consistent relationships between input and output parameters. Discovering the relationships, or namely compact laws, in a data set is of great interest in many fields, such as physics, chemistry, and finance. While data discovery has made great progress in practice thanks to the success of machine learning in recent years, the development of analytical approaches in finding the theory behind the data is relatively slow. In this paper, we develop an innovative approach in discovering compact laws from a data set. By proposing a novel algebraic equation formulation, we convert the problem of deriving meaning from data into formulating a linear algebra model and searching for relationships that fit the data. Rigorous proof is presented in validating the approach. The algebraic formulation allows the search of equation candidates in an explicit mathematical manner. Searching algorithms are also proposed for finding the governing equations with improved efficiency. For a certain type of compact theory, our approach assures convergence and the discovery is computationally efficient and mathematically precise.
Tasks
Published	2017-06-16
URL	http://arxiv.org/abs/1706.05123v1
PDF	http://arxiv.org/pdf/1706.05123v1.pdf
PWC	https://paperswithcode.com/paper/deriving-compact-laws-based-on-algebraic
Repo
Framework

Multi-Task Learning Using Neighborhood Kernels


Title	Multi-Task Learning Using Neighborhood Kernels
Authors	Niloofar Yousefi, Cong Li, Mansooreh Mollaghasemi, Georgios Anagnostopoulos, Michael Georgiopoulos
Abstract	This paper introduces a new and effective algorithm for learning kernels in a Multi-Task Learning (MTL) setting. Although, we consider a MTL scenario here, our approach can be easily applied to standard single task learning, as well. As shown by our empirical results, our algorithm consistently outperforms the traditional kernel learning algorithms such as uniform combination solution, convex combinations of base kernels as well as some kernel alignment-based models, which have been proven to give promising results in the past. We present a Rademacher complexity bound based on which a new Multi-Task Multiple Kernel Learning (MT-MKL) model is derived. In particular, we propose a Support Vector Machine-regularized model in which, for each task, an optimal kernel is learned based on a neighborhood-defining kernel that is not restricted to be positive semi-definite. Comparative experimental results are showcased that underline the merits of our neighborhood-defining framework in both classification and regression problems.
Tasks	Multi-Task Learning
Published	2017-07-11
URL	http://arxiv.org/abs/1707.03426v1
PDF	http://arxiv.org/pdf/1707.03426v1.pdf
PWC	https://paperswithcode.com/paper/multi-task-learning-using-neighborhood
Repo
Framework

Dense scale selection over space, time and space-time


Title	Dense scale selection over space, time and space-time
Authors	Tony Lindeberg
Abstract	Scale selection methods based on local extrema over scale of scale-normalized derivatives have been primarily developed to be applied sparsely — at image points where the magnitude of a scale-normalized differential expression additionally assumes local extrema over the domain where the data are defined. This paper presents a methodology for performing dense scale selection, so that hypotheses about local characteristic scales in images, temporal signals and video can be computed at every image point and every time moment. A critical problem when designing mechanisms for dense scale selection is that the scale at which scale-normalized differential entities assume local extrema over scale can be strongly dependent on the local order of the locally dominant differential structure. To address this problem, we propose a methodology where local extrema over scale are detected of a quasi quadrature measure involving scale-space derivatives up to order two and propose two independent mechanisms to reduce the phase dependency of the local scale estimates by: (i) introducing a second layer of post-smoothing prior to the detection of local extrema over scale and (ii) performing local phase compensation based on a model of the phase dependency of the local scale estimates depending on the relative strengths between first- vs. second-order differential structure. This general methodology is applied over three types of domains: (i) spatial images, (ii) temporal signals and (iii) spatio-temporal video. Experiments show that the proposed methodology leads to intuitively reasonable results with local scale estimates that reflect variations in the characteristic scales of locally dominant structures over space and time.
Tasks
Published	2017-09-25
URL	http://arxiv.org/abs/1709.08603v5
PDF	http://arxiv.org/pdf/1709.08603v5.pdf
PWC	https://paperswithcode.com/paper/dense-scale-selection-over-space-time-and
Repo
Framework

Reconstruction of~3-D Rigid Smooth Curves Moving Free when Two Traceable Points Only are Available


Title	Reconstruction of~3-D Rigid Smooth Curves Moving Free when Two Traceable Points Only are Available
Authors	Mieczysław A. Kłopotek
Abstract	This paper extends previous research in that sense that for orthogonal projections of rigid smooth (true-3D) curves moving totally free it reduces the number of required traceable points to two only (the best results known so far to the author are 3 points from free motion and 2 for motion restricted to rotation around a fixed direction and and 2 for motion restricted to influence of a homogeneous force field). The method used is exploitation of information on tangential projections. It discusses also possibility of simplification of reconstruction of flat curves moving free for prospective projections.
Tasks
Published	2017-04-11
URL	http://arxiv.org/abs/1704.03375v1
PDF	http://arxiv.org/pdf/1704.03375v1.pdf
PWC	https://paperswithcode.com/paper/reconstruction-of3-d-rigid-smooth-curves
Repo
Framework

On Pre-Trained Image Features and Synthetic Images for Deep Learning


Title	On Pre-Trained Image Features and Synthetic Images for Deep Learning
Authors	Stefan Hinterstoisser, Vincent Lepetit, Paul Wohlhart, Kurt Konolige
Abstract	Deep Learning methods usually require huge amounts of training data to perform at their full potential, and often require expensive manual labeling. Using synthetic images is therefore very attractive to train object detectors, as the labeling comes for free, and several approaches have been proposed to combine synthetic and real images for training. In this paper, we show that a simple trick is sufficient to train very effectively modern object detectors with synthetic images only: We freeze the layers responsible for feature extraction to generic layers pre-trained on real images, and train only the remaining layers with plain OpenGL rendering. Our experiments with very recent deep architectures for object recognition (Faster-RCNN, R-FCN, Mask-RCNN) and image feature extractors (InceptionResnet and Resnet) show this simple approach performs surprisingly well.
Tasks	Object Recognition
Published	2017-10-29
URL	http://arxiv.org/abs/1710.10710v2
PDF	http://arxiv.org/pdf/1710.10710v2.pdf
PWC	https://paperswithcode.com/paper/on-pre-trained-image-features-and-synthetic
Repo
Framework

Two Birds with One Stone: Transforming and Generating Facial Images with Iterative GAN


Title	Two Birds with One Stone: Transforming and Generating Facial Images with Iterative GAN
Authors	Dan Ma, Bin Liu, Zhao Kang, Jiayu Zhou, Jianke Zhu, Zenglin Xu
Abstract	Generating high fidelity identity-preserving faces with different facial attributes has a wide range of applications. Although a number of generative models have been developed to tackle this problem, there is still much room for further improvement.In paticular, the current solutions usually ignore the perceptual information of images, which we argue that it benefits the output of a high-quality image while preserving the identity information, especially in facial attributes learning area.To this end, we propose to train GAN iteratively via regularizing the min-max process with an integrated loss, which includes not only the per-pixel loss but also the perceptual loss. In contrast to the existing methods only deal with either image generation or transformation, our proposed iterative architecture can achieve both of them. Experiments on the multi-label facial dataset CelebA demonstrate that the proposed model has excellent performance on recognizing multiple attributes, generating a high-quality image, and transforming image with controllable attributes.
Tasks	Image Generation
Published	2017-11-16
URL	http://arxiv.org/abs/1711.06078v2
PDF	http://arxiv.org/pdf/1711.06078v2.pdf
PWC	https://paperswithcode.com/paper/two-birds-with-one-stone-transforming-and
Repo
Framework

Distance-to-Mean Continuous Conditional Random Fields to Enhance Prediction Problem in Traffic Flow Data


Title	Distance-to-Mean Continuous Conditional Random Fields to Enhance Prediction Problem in Traffic Flow Data
Authors	Sumarsih Condroayu Purbarani, Hadaiq Rolis Sanabila, Wisnu Jatmiko
Abstract	The increase of vehicle in highways may cause traffic congestion as well as in the normal roadways. Predicting the traffic flow in highways especially, is demanded to solve this congestion problem. Predictions on time-series multivariate data, such as in the traffic flow dataset, have been largely accomplished through various approaches. The approach with conventional prediction algorithms, such as with Support Vector Machine (SVM), is only capable of accommodating predictions that are independent in each time unit. Hence, the sequential relationships in this time series data is hardly explored. Continuous Conditional Random Field (CCRF) is one of Probabilistic Graphical Model (PGM) algorithms which can accommodate this problem. The neighboring aspects of sequential data such as in the time series data can be expressed by CCRF so that its predictions are more reliable. In this article, a novel approach called DM-CCRF is adopted by modifying the CCRF prediction algorithm to strengthen the probability of the predictions made by the baseline regressor. The result shows that DM-CCRF is superior in performance compared to CCRF. This is validated by the error decrease of the baseline up to 9% significance. This is twice the standard CCRF performance which can only decrease baseline error by 4.582% at most.
Tasks	Time Series
Published	2017-07-11
URL	http://arxiv.org/abs/1707.03110v1
PDF	http://arxiv.org/pdf/1707.03110v1.pdf
PWC	https://paperswithcode.com/paper/distance-to-mean-continuous-conditional
Repo
Framework

Note on Representing attribute reduction and concepts in concepts lattice using graphs


Title	Note on Representing attribute reduction and concepts in concepts lattice using graphs
Authors	Jan Konecny
Abstract	Mao H. (2017, Representing attribute reduction and concepts in concept lattice using graphs. Soft Computing 21(24):7293–7311) claims to make contributions to the study of reduction of attributes in concept lattices by using graph theory. We show that her results are either trivial or already well-known and all three algorithms proposed in the paper are incorrect.
Tasks
Published	2017-11-15
URL	http://arxiv.org/abs/1711.05509v2
PDF	http://arxiv.org/pdf/1711.05509v2.pdf
PWC	https://paperswithcode.com/paper/note-on-representing-attribute-reduction-and
Repo
Framework

Underestimate Sequences via Quadratic Averaging


Title	Underestimate Sequences via Quadratic Averaging
Authors	Chenxin Ma, Naga Venkata C. Gudapati, Majid Jahani, Rachael Tappenden, Martin Takáč
Abstract	In this work we introduce the concept of an Underestimate Sequence (UES), which is a natural extension of Nesterov’s estimate sequence. Our definition of a UES utilizes three sequences, one of which is a lower bound (or under-estimator) of the objective function. The question of how to construct an appropriate sequence of lower bounds is also addressed, and we present lower bounds for strongly convex smooth functions and for strongly convex composite functions, which adhere to the UES framework. Further, we propose several first order methods for minimizing strongly convex functions in both the smooth and composite cases. The algorithms, based on efficiently updating lower bounds on the objective functions, have natural stopping conditions, which provides the user with a certificate of optimality. Convergence of all algorithms is guaranteed through the UES framework, and we show that all presented algorithms converge linearly, with the accelerated variants enjoying the optimal linear rate of convergence.
Tasks
Published	2017-10-10
URL	http://arxiv.org/abs/1710.03695v1
PDF	http://arxiv.org/pdf/1710.03695v1.pdf
PWC	https://paperswithcode.com/paper/underestimate-sequences-via-quadratic
Repo
Framework

Learning Local Shape Descriptors from Part Correspondences With Multi-view Convolutional Networks


Title	Learning Local Shape Descriptors from Part Correspondences With Multi-view Convolutional Networks
Authors	Haibin Huang, Evangelos Kalogerakis, Siddhartha Chaudhuri, Duygu Ceylan, Vladimir G. Kim, Ersin Yumer
Abstract	We present a new local descriptor for 3D shapes, directly applicable to a wide range of shape analysis problems such as point correspondences, semantic segmentation, affordance prediction, and shape-to-scan matching. The descriptor is produced by a convolutional network that is trained to embed geometrically and semantically similar points close to one another in descriptor space. The network processes surface neighborhoods around points on a shape that are captured at multiple scales by a succession of progressively zoomed out views, taken from carefully selected camera positions. We leverage two extremely large sources of data to train our network. First, since our network processes rendered views in the form of 2D images, we repurpose architectures pre-trained on massive image datasets. Second, we automatically generate a synthetic dense point correspondence dataset by non-rigid alignment of corresponding shape parts in a large collection of segmented 3D models. As a result of these design choices, our network effectively encodes multi-scale local context and fine-grained surface detail. Our network can be trained to produce either category-specific descriptors or more generic descriptors by learning from multiple shape categories. Once trained, at test time, the network extracts local descriptors for shapes without requiring any part segmentation as input. Our method can produce effective local descriptors even for shapes whose category is unknown or different from the ones used while training. We demonstrate through several experiments that our learned local descriptors are more discriminative compared to state of the art alternatives, and are effective in a variety of shape analysis applications.
Tasks	Semantic Segmentation
Published	2017-06-14
URL	http://arxiv.org/abs/1706.04496v2
PDF	http://arxiv.org/pdf/1706.04496v2.pdf
PWC	https://paperswithcode.com/paper/learning-local-shape-descriptors-from-part
Repo
Framework

Computation Error Analysis of Block Floating Point Arithmetic Oriented Convolution Neural Network Accelerator Design


Title	Computation Error Analysis of Block Floating Point Arithmetic Oriented Convolution Neural Network Accelerator Design
Authors	Zhourui Song, Zhenyu Liu, Dongsheng Wang
Abstract	The heavy burdens of computation and off-chip traffic impede deploying the large scale convolution neural network on embedded platforms. As CNN is attributed to the strong endurance to computation errors, employing block floating point (BFP) arithmetics in CNN accelerators could save the hardware cost and data traffics efficiently, while maintaining the classification accuracy. In this paper, we verify the effects of word width definitions in BFP to the CNN performance without retraining. Several typical CNN models, including VGG16, ResNet-18, ResNet-50 and GoogLeNet, were tested in this paper. Experiments revealed that 8-bit mantissa, including sign bit, in BFP representation merely induced less than 0.3% accuracy loss. In addition, we investigate the computational errors in theory and develop the noise-to-signal ratio (NSR) upper bound, which provides the promising guidance for BFP based CNN engine design.
Tasks
Published	2017-09-22
URL	http://arxiv.org/abs/1709.07776v2
PDF	http://arxiv.org/pdf/1709.07776v2.pdf
PWC	https://paperswithcode.com/paper/computation-error-analysis-of-block-floating
Repo
Framework

Computer-aided implant design for the restoration of cranial defects


Title	Computer-aided implant design for the restoration of cranial defects
Authors	Xiaojun Chen, Lu Xu, Xing Li, Jan Egger
Abstract	Patient-specific cranial implants are important and necessary in the surgery of cranial defect restoration. However, traditional methods of manual design of cranial implants are complicated and time-consuming. Our purpose is to develop a novel software named EasyCrania to design the cranial implants conveniently and efficiently. The process can be divided into five steps, which are mirroring model, clipping surface, surface fitting, the generation of the initial implant and the generation of the final implant. The main concept of our method is to use the geometry information of the mirrored model as the base to generate the final implant. The comparative studies demonstrated that the EasyCrania can improve the efficiency of cranial implant design significantly. And, the intra- and inter-rater reliability of the software were stable, which were 87.07+/-1.6% and 87.73+/-1.4% respectively.
Tasks
Published	2017-06-23
URL	http://arxiv.org/abs/1706.07649v1
PDF	http://arxiv.org/pdf/1706.07649v1.pdf
PWC	https://paperswithcode.com/paper/computer-aided-implant-design-for-the
Repo
Framework