July 27, 2019

3504 words 17 mins read

Paper Group ANR 719

Online Video Deblurring via Dynamic Temporal Blending Network. The Cafe Wall Illusion: Local and Global Perception from multiple scale to multiscale. Adversarial Training for Disease Prediction from Electronic Health Records with Missing Data. Learning what to look in chest X-rays with a recurrent visual attention model. Scalable Joint Models for R …

Online Video Deblurring via Dynamic Temporal Blending Network


Title	Online Video Deblurring via Dynamic Temporal Blending Network
Authors	Tae Hyun Kim, Kyoung Mu Lee, Bernhard Schölkopf, Michael Hirsch
Abstract	State-of-the-art video deblurring methods are capable of removing non-uniform blur caused by unwanted camera shake and/or object motion in dynamic scenes. However, most existing methods are based on batch processing and thus need access to all recorded frames, rendering them computationally demanding and time consuming and thus limiting their practical use. In contrast, we propose an online (sequential) video deblurring method based on a spatio-temporal recurrent network that allows for real-time performance. In particular, we introduce a novel architecture which extends the receptive field while keeping the overall size of the network small to enable fast execution. In doing so, our network is able to remove even large blur caused by strong camera shake and/or fast moving objects. Furthermore, we propose a novel network layer that enforces temporal consistency between consecutive frames by dynamic temporal blending which compares and adaptively (at test time) shares features obtained at different time steps. We show the superiority of the proposed method in an extensive experimental evaluation.
Tasks	Deblurring
Published	2017-04-11
URL	http://arxiv.org/abs/1704.03285v1
PDF	http://arxiv.org/pdf/1704.03285v1.pdf
PWC	https://paperswithcode.com/paper/online-video-deblurring-via-dynamic-temporal
Repo
Framework

The Cafe Wall Illusion: Local and Global Perception from multiple scale to multiscale


Title	The Cafe Wall Illusion: Local and Global Perception from multiple scale to multiscale
Authors	Nasim Nematzadeh, David M. W. Powers
Abstract	Geometrical illusions are a subclass of optical illusions in which the geometrical characteristics of patterns such as orientations and angles are distorted and misperceived as the result of low- to high-level retinal/cortical processing. Modelling the detection of tilt in these illusions and their strengths as they are perceived is a challenging task computationally and leads to development of techniques that match with human performance. In this study, we present a predictive and quantitative approach for modeling foveal and peripheral vision in the induced tilt in Caf'e Wall illusion in which parallel mortar lines between shifted rows of black and white tiles appear to converge and diverge. A bioderived filtering model for the responses of retinal/cortical simple cells to the stimulus using Difference of Gaussians is utilized with an analytic processing pipeline introduced in our previous studies to quantify the angle of tilt in the model. Here we have considered visual characteristics of foveal and peripheral vision in the perceived tilt in the pattern to predict different degrees of tilt in different areas of the fovea and periphery as the eye saccades to different parts of the image. The tilt analysis results from several sampling sizes and aspect ratios, modelling variant foveal views are used from our previous investigations on the local tilt, and we specifically investigate in this work, different configurations of the whole pattern modelling variant Gestalt views across multiple scales in order to provide confidence intervals around the predicted tilts. The foveal sample sets are verified and quantified using two different sampling methods. We present here a precise and quantified comparison contrasting local tilt detection in the foveal sets with a global average across all of the Caf'e Wall configurations tested in this work.
Tasks
Published	2017-09-17
URL	http://arxiv.org/abs/1710.01215v1
PDF	http://arxiv.org/pdf/1710.01215v1.pdf
PWC	https://paperswithcode.com/paper/the-cafe-wall-illusion-local-and-global
Repo
Framework

Adversarial Training for Disease Prediction from Electronic Health Records with Missing Data


Title	Adversarial Training for Disease Prediction from Electronic Health Records with Missing Data
Authors	Uiwon Hwang, Sungwoon Choi, Han-Byoel Lee, Sungroh Yoon
Abstract	Electronic health records (EHRs) have contributed to the computerization of patient records and can thus be used not only for efficient and systematic medical services, but also for research on biomedical data science. However, there are many missing values in EHRs when provided in matrix form, which is an important issue in many biomedical EHR applications. In this paper, we propose a two-stage framework that includes missing data imputation and disease prediction to address the missing data problem in EHRs. We compared the disease prediction performance of generative adversarial networks (GANs) and conventional learning algorithms in combination with missing data prediction methods. As a result, we obtained a level of accuracy of 0.9777, sensitivity of 0.9521, specificity of 0.9925, area under the receiver operating characteristic curve (AUC-ROC) of 0.9889, and F-score of 0.9688 with a stacked autoencoder as the missing data prediction method and an auxiliary classifier GAN (AC-GAN) as the disease prediction method. The comparison results show that a combination of a stacked autoencoder and an AC-GAN significantly outperforms other existing approaches. Our results suggest that the proposed framework is more robust for disease prediction from EHRs with missing data.
Tasks	Disease Prediction, Imputation
Published	2017-11-11
URL	http://arxiv.org/abs/1711.04126v4
PDF	http://arxiv.org/pdf/1711.04126v4.pdf
PWC	https://paperswithcode.com/paper/adversarial-training-for-disease-prediction
Repo
Framework

Learning what to look in chest X-rays with a recurrent visual attention model


Title	Learning what to look in chest X-rays with a recurrent visual attention model
Authors	Petros-Pavlos Ypsilantis, Giovanni Montana
Abstract	X-rays are commonly performed imaging tests that use small amounts of radiation to produce pictures of the organs, tissues, and bones of the body. X-rays of the chest are used to detect abnormalities or diseases of the airways, blood vessels, bones, heart, and lungs. In this work we present a stochastic attention-based model that is capable of learning what regions within a chest X-ray scan should be visually explored in order to conclude that the scan contains a specific radiological abnormality. The proposed model is a recurrent neural network (RNN) that learns to sequentially sample the entire X-ray and focus only on informative areas that are likely to contain the relevant information. We report on experiments carried out with more than $100,000$ X-rays containing enlarged hearts or medical devices. The model has been trained using reinforcement learning methods to learn task-specific policies.
Tasks
Published	2017-01-23
URL	http://arxiv.org/abs/1701.06452v1
PDF	http://arxiv.org/pdf/1701.06452v1.pdf
PWC	https://paperswithcode.com/paper/learning-what-to-look-in-chest-x-rays-with-a
Repo
Framework

Scalable Joint Models for Reliable Uncertainty-Aware Event Prediction


Title	Scalable Joint Models for Reliable Uncertainty-Aware Event Prediction
Authors	Hossein Soleimani, James Hensman, Suchi Saria
Abstract	Missing data and noisy observations pose significant challenges for reliably predicting events from irregularly sampled multivariate time series (longitudinal) data. Imputation methods, which are typically used for completing the data prior to event prediction, lack a principled mechanism to account for the uncertainty due to missingness. Alternatively, state-of-the-art joint modeling techniques can be used for jointly modeling the longitudinal and event data and compute event probabilities conditioned on the longitudinal observations. These approaches, however, make strong parametric assumptions and do not easily scale to multivariate signals with many observations. Our proposed approach consists of several key innovations. First, we develop a flexible and scalable joint model based upon sparse multiple-output Gaussian processes. Unlike state-of-the-art joint models, the proposed model can explain highly challenging structure including non-Gaussian noise while scaling to large data. Second, we derive an optimal policy for predicting events using the distribution of the event occurrence estimated by the joint model. The derived policy trades-off the cost of a delayed detection versus incorrect assessments and abstains from making decisions when the estimated event probability does not satisfy the derived confidence criteria. Experiments on a large dataset show that the proposed framework significantly outperforms state-of-the-art techniques in event prediction.
Tasks	Gaussian Processes, Imputation, Time Series
Published	2017-08-16
URL	http://arxiv.org/abs/1708.04757v1
PDF	http://arxiv.org/pdf/1708.04757v1.pdf
PWC	https://paperswithcode.com/paper/scalable-joint-models-for-reliable
Repo
Framework

Fusing Bird View LIDAR Point Cloud and Front View Camera Image for Deep Object Detection


Title	Fusing Bird View LIDAR Point Cloud and Front View Camera Image for Deep Object Detection
Authors	Zining Wang, Wei Zhan, Masayoshi Tomizuka
Abstract	We propose a new method for fusing a LIDAR point cloud and camera-captured images in the deep convolutional neural network (CNN). The proposed method constructs a new layer called non-homogeneous pooling layer to transform features between bird view map and front view map. The sparse LIDAR point cloud is used to construct the mapping between the two maps. The pooling layer allows efficient fusion of the bird view and front view features at any stage of the network. This is favorable for the 3D-object detection using camera-LIDAR fusion in autonomous driving scenarios. A corresponding deep CNN is designed and tested on the KITTI bird view object detection dataset, which produces 3D bounding boxes from the bird view map. The fusion method shows particular benefit for detection of pedestrians in the bird view compared to other fusion-based object detection networks.
Tasks	3D Object Detection, Autonomous Driving, Object Detection
Published	2017-11-17
URL	http://arxiv.org/abs/1711.06703v3
PDF	http://arxiv.org/pdf/1711.06703v3.pdf
PWC	https://paperswithcode.com/paper/fusing-bird-view-lidar-point-cloud-and-front
Repo
Framework

Universal 3D Wearable Fingerprint Targets: Advancing Fingerprint Reader Evaluations


Title	Universal 3D Wearable Fingerprint Targets: Advancing Fingerprint Reader Evaluations
Authors	Joshua J. Engelsma, Sunpreet S. Arora, Anil K. Jain, Nicholas G. Paulter Jr
Abstract	We present the design and manufacturing of high fidelity universal 3D fingerprint targets, which can be imaged on a variety of fingerprint sensing technologies, namely capacitive, contact-optical, and contactless-optical. Universal 3D fingerprint targets enable, for the first time, not only a repeatable and controlled evaluation of fingerprint readers, but also the ability to conduct fingerprint reader interoperability studies. Fingerprint reader interoperability refers to how robust fingerprint recognition systems are to variations in the images acquired by different types of fingerprint readers. To build universal 3D fingerprint targets, we adopt a molding and casting framework consisting of (i) digital mapping of fingerprint images to a negative mold, (ii) CAD modeling a scaffolding system to hold the negative mold, (iii) fabricating the mold and scaffolding system with a high resolution 3D printer, (iv) producing or mixing a material with similar electrical, optical, and mechanical properties to that of the human finger, and (v) fabricating a 3D fingerprint target using controlled casting. Our experiments conducted with PIV and Appendix F certified optical (contact and contactless) and capacitive fingerprint readers demonstrate the usefulness of universal 3D fingerprint targets for controlled and repeatable fingerprint reader evaluations and also fingerprint reader interoperability studies.
Tasks
Published	2017-05-22
URL	http://arxiv.org/abs/1705.07972v1
PDF	http://arxiv.org/pdf/1705.07972v1.pdf
PWC	https://paperswithcode.com/paper/universal-3d-wearable-fingerprint-targets
Repo
Framework

Deep Q-Learning for Self-Organizing Networks Fault Management and Radio Performance Improvement


Title	Deep Q-Learning for Self-Organizing Networks Fault Management and Radio Performance Improvement
Authors	Faris B. Mismar, Brian L. Evans
Abstract	We propose an algorithm to automate fault management in an outdoor cellular network using deep reinforcement learning (RL) against wireless impairments. This algorithm enables the cellular network cluster to self-heal by allowing RL to learn how to improve the downlink signal to interference plus noise ratio through exploration and exploitation of various alarm corrective actions. The main contributions of this paper are to 1) introduce a deep RL-based fault handling algorithm which self-organizing networks can implement in a polynomial runtime and 2) show that this fault management method can improve the radio link performance in a realistic network setup. Simulation results show that our proposed algorithm learns an action sequence to clear alarms and improve the performance in the cellular cluster better than existing algorithms, even against the randomness of the network fault occurrences and user movements.
Tasks	Q-Learning
Published	2017-07-10
URL	http://arxiv.org/abs/1707.02329v6
PDF	http://arxiv.org/pdf/1707.02329v6.pdf
PWC	https://paperswithcode.com/paper/deep-q-learning-for-self-organizing-networks
Repo
Framework

SQG-Differential Evolution for difficult optimization problems under a tight function evaluation budget


Title	SQG-Differential Evolution for difficult optimization problems under a tight function evaluation budget
Authors	Ramses Sala, Niccolo Baldanzini, Marco Pierini
Abstract	In the context of industrial engineering, it is important to integrate efficient computational optimization methods in the product development process. Some of the most challenging simulation-based engineering design optimization problems are characterized by: a large number of design variables, the absence of analytical gradients, highly non-linear objectives and a limited function evaluation budget. Although a huge variety of different optimization algorithms is available, the development and selection of efficient algorithms for problems with these industrial relevant characteristics, remains a challenge. In this communication, a hybrid variant of Differential Evolution (DE) is introduced which combines aspects of Stochastic Quasi-Gradient (SQG) methods within the framework of DE, in order to improve optimization efficiency on problems with the previously mentioned characteristics. The performance of the resulting derivative-free algorithm is compared with other state-of-the-art DE variants on 25 commonly used benchmark functions, under tight function evaluation budget constraints of 1000 evaluations. The experimental results indicate that the new algorithm performs excellent on the ‘difficult’ (high dimensional, multi-modal, inseparable) test functions. The operations used in the proposed mutation scheme, are computationally inexpensive, and can be easily implemented in existing differential evolution variants or other population-based optimization algorithms by a few lines of program code as an non-invasive optional setting. Besides the applicability of the presented algorithm by itself, the described concepts can serve as a useful and interesting addition to the algorithmic operators in the frameworks of heuristics and evolutionary optimization and computing.
Tasks
Published	2017-10-18
URL	http://arxiv.org/abs/1710.06770v2
PDF	http://arxiv.org/pdf/1710.06770v2.pdf
PWC	https://paperswithcode.com/paper/sqg-differential-evolution-for-difficult
Repo
Framework

DeepCodec: Adaptive Sensing and Recovery via Deep Convolutional Neural Networks


Title	DeepCodec: Adaptive Sensing and Recovery via Deep Convolutional Neural Networks
Authors	Ali Mousavi, Gautam Dasarathy, Richard G. Baraniuk
Abstract	In this paper we develop a novel computational sensing framework for sensing and recovering structured signals. When trained on a set of representative signals, our framework learns to take undersampled measurements and recover signals from them using a deep convolutional neural network. In other words, it learns a transformation from the original signals to a near-optimal number of undersampled measurements and the inverse transformation from measurements to signals. This is in contrast to traditional compressive sensing (CS) systems that use random linear measurements and convex optimization or iterative algorithms for signal recovery. We compare our new framework with $\ell_1$-minimization from the phase transition point of view and demonstrate that it outperforms $\ell_1$-minimization in the regions of phase transition plot where $\ell_1$-minimization cannot recover the exact solution. In addition, we experimentally demonstrate how learning measurements enhances the overall recovery performance, speeds up training of recovery framework, and leads to having fewer parameters to learn.
Tasks	Compressive Sensing
Published	2017-07-11
URL	http://arxiv.org/abs/1707.03386v1
PDF	http://arxiv.org/pdf/1707.03386v1.pdf
PWC	https://paperswithcode.com/paper/deepcodec-adaptive-sensing-and-recovery-via
Repo
Framework

RIDDLE: Race and ethnicity Imputation from Disease history with Deep LEarning


Title	RIDDLE: Race and ethnicity Imputation from Disease history with Deep LEarning
Authors	Ji-Sung Kim, Xin Gao, Andrey Rzhetsky
Abstract	Anonymized electronic medical records are an increasingly popular source of research data. However, these datasets often lack race and ethnicity information. This creates problems for researchers modeling human disease, as race and ethnicity are powerful confounders for many health exposures and treatment outcomes; race and ethnicity are closely linked to population-specific genetic variation. We showed that deep neural networks generate more accurate estimates for missing racial and ethnic information than competing methods (e.g., logistic regression, random forest). RIDDLE yielded significantly better classification performance across all metrics that were considered: accuracy, cross-entropy loss (error), and area under the curve for receiver operating characteristic plots (all $p < 10^{-6}$). We made specific efforts to interpret the trained neural network models to identify, quantify, and visualize medical features which are predictive of race and ethnicity. We used these characterizations of informative features to perform a systematic comparison of differential disease patterns by race and ethnicity. The fact that clinical histories are informative for imputing race and ethnicity could reflect (1) a skewed distribution of blue- and white-collar professions across racial and ethnic groups, (2) uneven accessibility and subjective importance of prophylactic health, (3) possible variation in lifestyle, such as dietary habits, and (4) differences in background genetic variation which predispose to diseases.
Tasks	Imputation
Published	2017-07-06
URL	http://arxiv.org/abs/1707.01623v2
PDF	http://arxiv.org/pdf/1707.01623v2.pdf
PWC	https://paperswithcode.com/paper/riddle-race-and-ethnicity-imputation-from
Repo
Framework

Automatic segmentation of MR brain images with a convolutional neural network


Title	Automatic segmentation of MR brain images with a convolutional neural network
Authors	Pim Moeskops, Max A. Viergever, Adriënne M. Mendrik, Linda S. de Vries, Manon J. N. L. Benders, Ivana Išgum
Abstract	Automatic segmentation in MR brain images is important for quantitative analysis in large-scale studies with images acquired at all ages. This paper presents a method for the automatic segmentation of MR brain images into a number of tissue classes using a convolutional neural network. To ensure that the method obtains accurate segmentation details as well as spatial consistency, the network uses multiple patch sizes and multiple convolution kernel sizes to acquire multi-scale information about each voxel. The method is not dependent on explicit features, but learns to recognise the information that is important for the classification based on training data. The method requires a single anatomical MR image only. The segmentation method is applied to five different data sets: coronal T2-weighted images of preterm infants acquired at 30 weeks postmenstrual age (PMA) and 40 weeks PMA, axial T2- weighted images of preterm infants acquired at 40 weeks PMA, axial T1-weighted images of ageing adults acquired at an average age of 70 years, and T1-weighted images of young adults acquired at an average age of 23 years. The method obtained the following average Dice coefficients over all segmented tissue classes for each data set, respectively: 0.87, 0.82, 0.84, 0.86 and 0.91. The results demonstrate that the method obtains accurate segmentations in all five sets, and hence demonstrates its robustness to differences in age and acquisition protocol.
Tasks
Published	2017-04-11
URL	http://arxiv.org/abs/1704.03295v1
PDF	http://arxiv.org/pdf/1704.03295v1.pdf
PWC	https://paperswithcode.com/paper/automatic-segmentation-of-mr-brain-images
Repo
Framework

Dimensionality reduction with missing values imputation


Title	Dimensionality reduction with missing values imputation
Authors	Rania Mkhinini Gahar, Olfa Arfaoui, Minyar Sassi Hidri, Nejib Ben-Hadj Alouane
Abstract	In this study, we propose a new statical approach for high-dimensionality reduction of heterogenous data that limits the curse of dimensionality and deals with missing values. To handle these latter, we propose to use the Random Forest imputation’s method. The main purpose here is to extract useful information and so reducing the search space to facilitate the data exploration process. Several illustrative numeric examples, using data coming from publicly available machine learning repositories are also included. The experimental component of the study shows the efficiency of the proposed analytical approach.
Tasks	Dimensionality Reduction, Imputation
Published	2017-07-02
URL	http://arxiv.org/abs/1707.00351v1
PDF	http://arxiv.org/pdf/1707.00351v1.pdf
PWC	https://paperswithcode.com/paper/dimensionality-reduction-with-missing-values
Repo
Framework

3D-SSD: Learning Hierarchical Features from RGB-D Images for Amodal 3D Object Detection


Title	3D-SSD: Learning Hierarchical Features from RGB-D Images for Amodal 3D Object Detection
Authors	Qianhui Luo, Huifang Ma, Yue Wang, Li Tang, Rong Xiong
Abstract	This paper aims at developing a faster and a more accurate solution to the amodal 3D object detection problem for indoor scenes. It is achieved through a novel neural network that takes a pair of RGB-D images as the input and delivers oriented 3D bounding boxes as the output. The network, named 3D-SSD, composed of two parts: hierarchical feature fusion and multi-layer prediction. The hierarchical feature fusion combines appearance and geometric features from RGB-D images while the multi-layer prediction utilizes multi-scale features for object detection. As a result, the network can exploit 2.5D representations in a synergetic way to improve the accuracy and efficiency. The issue of object sizes is addressed by attaching a set of 3D anchor boxes with varying sizes to every location of the prediction layers. At the end stage, the category scores for 3D anchor boxes are generated with adjusted positions, sizes and orientations respectively, leading to the final detections using non-maximum suppression. In the training phase, the positive samples are identified with the aid of 2D ground truth to avoid the noisy estimation of depth from raw data, which guide to a better converged model. Experiments performed on the challenging SUN RGB-D dataset show that our algorithm outperforms the state-of-the-art Deep Sliding Shape by 10.2% mAP and 88x faster. Further, experiments also suggest our approach achieves comparable accuracy and is 386x faster than the state-of-art method on the NYUv2 dataset even with a smaller input image size.
Tasks	3D Object Detection, Object Detection
Published	2017-11-01
URL	http://arxiv.org/abs/1711.00238v2
PDF	http://arxiv.org/pdf/1711.00238v2.pdf
PWC	https://paperswithcode.com/paper/3d-ssd-learning-hierarchical-features-from
Repo
Framework

In search of inliers: 3d correspondence by local and global voting


Title	In search of inliers: 3d correspondence by local and global voting
Authors	Anders Glent Buch, Yang Yang, Norbert Krüger, Henrik Gordon Petersen
Abstract	We present a method for finding correspondence between 3D models. From an initial set of feature correspondences, our method uses a fast voting scheme to separate the inliers from the outliers. The novelty of our method lies in the use of a combination of local and global constraints to determine if a vote should be cast. On a local scale, we use simple, low-level geometric invariants. On a global scale, we apply covariant constraints for finding compatible correspondences. We guide the sampling for collecting voters by downward dependencies on previous voting stages. All of this together results in an accurate matching procedure. We evaluate our algorithm by controlled and comparative testing on different datasets, giving superior performance compared to state of the art methods. In a final experiment, we apply our method for 3D object detection, showing potential use of our method within higher-level vision.
Tasks	3D Object Detection, Object Detection
Published	2017-08-23
URL	http://arxiv.org/abs/1708.06966v1
PDF	http://arxiv.org/pdf/1708.06966v1.pdf
PWC	https://paperswithcode.com/paper/in-search-of-inliers-3d-correspondence-by
Repo
Framework