January 30, 2020

3115 words 15 mins read

Paper Group ANR 345

A Discrete Empirical Interpolation Method for Interpretable Immersion and Embedding of Nonlinear Manifolds. Object Segmentation using Pixel-wise Adversarial Loss. AdvSPADE: Realistic Unrestricted Attacks for Semantic Segmentation. Deep Value Model Predictive Control. StereoDRNet: Dilated Residual Stereo Net. Question Guided Modular Routing Networks …

A Discrete Empirical Interpolation Method for Interpretable Immersion and Embedding of Nonlinear Manifolds


Title	A Discrete Empirical Interpolation Method for Interpretable Immersion and Embedding of Nonlinear Manifolds
Authors	Samuel E. Otto, Clarence W. Rowley
Abstract	Manifold learning techniques seek to discover structure-preserving mappings of high-dimensional data into low-dimensional spaces. While the new sets of coordinates specified by these mappings can closely parameterize the data, they are generally complicated nonlinear functions of the original variables. This makes them difficult to interpret physically. Furthermore, in data-driven model reduction applications the governing equations may have structure that is destroyed by nonlinear mapping into coordinates on an inertial manifold, creating a computational bottleneck for simulations. Instead, we propose to identify a small collection of the original variables which are capable of uniquely determining all others either locally via immersion or globally via embedding of the underlying manifold. When the data lies on a low-dimensional subspace the existing discrete empirical interpolation method (DEIM) accomplishes this with recent variants employing greedy algorithms based on pivoted QR (PQR) factorizations. However, low-dimensional manifolds coming from a variety of applications, particularly from advection-dominated PDEs, do not lie in or near any low-dimensional subspace. Our proposed approach extends DEIM to data lying near nonlinear manifolds by applying a similar pivoted QR procedure simultaneously on collections of patches making up locally linear approximations of the manifold, resulting in a novel simultaneously pivoted QR (SimPQR) algorithm. The immersion provided by SimPQR can be extended to an embedding by applying SimPQR a second time to a modified collection of vectors. The SimPQR method for computing these `nonlinear DEIM’ (NLDEIM) coordinates is successfully applied to real-world data lying near an inertial manifold in a cylinder wake flow as well as data coming from a viscous Burgers equation with different initial conditions. \|
Tasks
Published	2019-05-18
URL	https://arxiv.org/abs/1905.07619v2
PDF	https://arxiv.org/pdf/1905.07619v2.pdf
PWC	https://paperswithcode.com/paper/a-discrete-empirical-interpolation-method-for
Repo
Framework

Object Segmentation using Pixel-wise Adversarial Loss


Title	Object Segmentation using Pixel-wise Adversarial Loss
Authors	Ricard Durall, Franz-Josef Pfreundt, Ullrich Köthe, Janis Keuper
Abstract	Recent deep learning based approaches have shown remarkable success on object segmentation tasks. However, there is still room for further improvement. Inspired by generative adversarial networks, we present a generic end-to-end adversarial approach, which can be combined with a wide range of existing semantic segmentation networks to improve their segmentation performance. The key element of our method is to replace the commonly used binary adversarial loss with a high resolution pixel-wise loss. In addition, we train our generator employing stochastic weight averaging fashion, which further enhances the predicted output label maps leading to state-of-the-art results. We show, that this combination of pixel-wise adversarial training and weight averaging leads to significant and consistent gains in segmentation performance, compared to the baseline models.
Tasks	Semantic Segmentation
Published	2019-09-23
URL	https://arxiv.org/abs/1909.10341v1
PDF	https://arxiv.org/pdf/1909.10341v1.pdf
PWC	https://paperswithcode.com/paper/190910341
Repo
Framework

AdvSPADE: Realistic Unrestricted Attacks for Semantic Segmentation


Title	AdvSPADE: Realistic Unrestricted Attacks for Semantic Segmentation
Authors	Guangyu Shen, Chengzhi Mao, Junfeng Yang, Baishakhi Ray
Abstract	Due to the inherent robustness of segmentation models, traditional norm-bounded attack methods show limited effect on such type of models. In this paper, we focus on generating unrestricted adversarial examples for semantic segmentation models. We demonstrate a simple and effective method to generate unrestricted adversarial examples using conditional generative adversarial networks (CGAN) without any hand-crafted metric. The na"ive implementation of CGAN, however, yields inferior image quality and low attack success rate. Instead, we leverage the SPADE (Spatially-adaptive denormalization) structure with an additional loss item to generate effective adversarial attacks in a single step. We validate our approach on the popular Cityscapes and ADE20K datasets, and demonstrate that our synthetic adversarial examples are not only realistic, but also improve the attack success rate by up to 41.0% compared with the state of the art adversarial attack methods including PGD.
Tasks	Adversarial Attack, Semantic Segmentation
Published	2019-10-06
URL	https://arxiv.org/abs/1910.02354v3
PDF	https://arxiv.org/pdf/1910.02354v3.pdf
PWC	https://paperswithcode.com/paper/unrestricted-adversarial-attacks-for-semantic-1
Repo
Framework

Deep Value Model Predictive Control


Title	Deep Value Model Predictive Control
Authors	Farbod Farshidian, David Hoeller, Marco Hutter
Abstract	In this paper, we introduce an actor-critic algorithm called Deep Value Model Predictive Control (DMPC), which combines model-based trajectory optimization with value function estimation. The DMPC actor is a Model Predictive Control (MPC) optimizer with an objective function defined in terms of a value function estimated by the critic. We show that our MPC actor is an importance sampler, which minimizes an upper bound of the cross-entropy to the state distribution of the optimal sampling policy. In our experiments with a Ballbot system, we show that our algorithm can work with sparse and binary reward signals to efficiently solve obstacle avoidance and target reaching tasks. Compared to previous work, we show that including the value function in the running cost of the trajectory optimizer speeds up the convergence. We also discuss the necessary strategies to robustify the algorithm in practice.
Tasks
Published	2019-10-08
URL	https://arxiv.org/abs/1910.03358v1
PDF	https://arxiv.org/pdf/1910.03358v1.pdf
PWC	https://paperswithcode.com/paper/deep-value-model-predictive-control
Repo
Framework

StereoDRNet: Dilated Residual Stereo Net


Title	StereoDRNet: Dilated Residual Stereo Net
Authors	Rohan Chabra, Julian Straub, Chris Sweeney, Richard Newcombe, Henry Fuchs
Abstract	We propose a system that uses a convolution neural network (CNN) to estimate depth from a stereo pair followed by volumetric fusion of the predicted depth maps to produce a 3D reconstruction of a scene. Our proposed depth refinement architecture, predicts view-consistent disparity and occlusion maps that helps the fusion system to produce geometrically consistent reconstructions. We utilize 3D dilated convolutions in our proposed cost filtering network that yields better filtering while almost halving the computational cost in comparison to state of the art cost filtering architectures.For feature extraction we use the Vortex Pooling architecture. The proposed method achieves state of the art results in KITTI 2012, KITTI 2015 and ETH 3D stereo benchmarks. Finally, we demonstrate that our system is able to produce high fidelity 3D scene reconstructions that outperforms the state of the art stereo system.
Tasks	3D Reconstruction
Published	2019-04-03
URL	https://arxiv.org/abs/1904.02251v3
PDF	https://arxiv.org/pdf/1904.02251v3.pdf
PWC	https://paperswithcode.com/paper/stereodrnet-dilated-residual-stereo-net
Repo
Framework

Question Guided Modular Routing Networks for Visual Question Answering


Title	Question Guided Modular Routing Networks for Visual Question Answering
Authors	Yanze Wu, Qiang Sun, Jianqi Ma, Bin Li, Yanwei Fu, Yao Peng, Xiangyang Xue
Abstract	Visual Question Answering (VQA) faces two major challenges: how to better fuse the visual and textual modalities and how to make the VQA model have the reasoning ability to answer more complex questions. In this paper, we address both challenges by proposing the novel Question Guided Modular Routing Networks (QGMRN). QGMRN can fuse the visual and textual modalities in multiple semantic levels which makes the fusion occur in a fine-grained way, it also can learn to reason by routing between the generic modules without additional supervision information or prior knowledge. The proposed QGMRN consists of three sub-networks: visual network, textual network and routing network. The routing network selectively executes each module in the visual network according to the pathway activated by the question features generated by the textual network. Experiments on the CLEVR dataset show that our model can outperform the state-of-the-art. Models and Codes will be released.
Tasks	Question Answering, Visual Question Answering
Published	2019-04-17
URL	https://arxiv.org/abs/1904.08324v2
PDF	https://arxiv.org/pdf/1904.08324v2.pdf
PWC	https://paperswithcode.com/paper/question-guided-modular-routing-networks-for
Repo
Framework

Trust and Cognitive Load During Human-Robot Interaction


Title	Trust and Cognitive Load During Human-Robot Interaction
Authors	Muneeb Imtiaz Ahmad, Jasmin Bernotat, Katrin Lohan, Friederike Eyssel
Abstract	This paper presents an exploratory study to understand the relationship between a humans’ cognitive load, trust, and anthropomorphism during human-robot interaction. To understand the relationship, we created a \say{Matching the Pair} game that participants could play collaboratively with one of two robot types, Husky or Pepper. The goal was to understand if humans would trust the robot as a teammate while being in the game-playing situation that demanded a high level of cognitive load. Using a humanoid vs. a technical robot, we also investigated the impact of physical anthropomorphism and we furthermore tested the impact of robot error rate on subsequent judgments and behavior. Our results showed that there was an inversely proportional relationship between trust and cognitive load, suggesting that as the amount of cognitive load increased in the participants, their ratings of trust decreased. We also found a triple interaction impact between robot-type, error-rate and participant’s ratings of trust. We found that participants perceived Pepper to be more trustworthy in comparison with the Husky robot after playing the game with both robots under high error-rate condition. On the contrary, Husky was perceived as more trustworthy than Pepper when it was depicted as featuring a low error-rate. Our results are interesting and call further investigation of the impact of physical anthropomorphism in combination with variable error-rates of the robot.
Tasks
Published	2019-09-11
URL	https://arxiv.org/abs/1909.05160v1
PDF	https://arxiv.org/pdf/1909.05160v1.pdf
PWC	https://paperswithcode.com/paper/trust-and-cognitive-load-during-human-robot
Repo
Framework

Gaussian Temporal Awareness Networks for Action Localization


Title	Gaussian Temporal Awareness Networks for Action Localization
Authors	Fuchen Long, Ting Yao, Zhaofan Qiu, Xinmei Tian, Jiebo Luo, Tao Mei
Abstract	Temporally localizing actions in a video is a fundamental challenge in video understanding. Most existing approaches have often drawn inspiration from image object detection and extended the advances, e.g., SSD and Faster R-CNN, to produce temporal locations of an action in a 1D sequence. Nevertheless, the results can suffer from robustness problem due to the design of predetermined temporal scales, which overlooks the temporal structure of an action and limits the utility on detecting actions with complex variations. In this paper, we propose to address the problem by introducing Gaussian kernels to dynamically optimize temporal scale of each action proposal. Specifically, we present Gaussian Temporal Awareness Networks (GTAN) — a new architecture that novelly integrates the exploitation of temporal structure into an one-stage action localization framework. Technically, GTAN models the temporal structure through learning a set of Gaussian kernels, each for a cell in the feature maps. Each Gaussian kernel corresponds to a particular interval of an action proposal and a mixture of Gaussian kernels could further characterize action proposals with various length. Moreover, the values in each Gaussian curve reflect the contextual contributions to the localization of an action proposal. Extensive experiments are conducted on both THUMOS14 and ActivityNet v1.3 datasets, and superior results are reported when comparing to state-of-the-art approaches. More remarkably, GTAN achieves 1.9% and 1.1% improvements in mAP on testing set of the two datasets.
Tasks	Action Localization, Object Detection, Video Understanding
Published	2019-09-09
URL	https://arxiv.org/abs/1909.03877v1
PDF	https://arxiv.org/pdf/1909.03877v1.pdf
PWC	https://paperswithcode.com/paper/gaussian-temporal-awareness-networks-for-1
Repo
Framework

A Scalable Framework for Acceleration of CNN Training on Deeply-Pipelined FPGA Clusters with Weight and Workload Balancing


Title	A Scalable Framework for Acceleration of CNN Training on Deeply-Pipelined FPGA Clusters with Weight and Workload Balancing
Authors	Tong Geng, Tianqi Wang, Ang Li, Xi Jin, Martin Herbordt
Abstract	Deep Neural Networks (DNNs) have revolutionized numerous applications, but the demand for ever more performance remains unabated. Scaling DNN computations to larger clusters is generally done by distributing tasks in batch mode using methods such as distributed synchronous SGD. Among the issues with this approach is that to make the distributed cluster work with high utilization, the workload distributed to each node must be large, which implies nontrivial growth in the SGD mini-batch size. In this paper, we propose a framework called FPDeep, which uses a hybrid of model and layer parallelism to configure distributed reconfigurable clusters to train DNNs. This approach has numerous benefits. First, the design does not suffer from batch size growth. Second, novel workload and weight partitioning leads to balanced loads of both among nodes. And third, the entire system is a fine-grained pipeline. This leads to high parallelism and utilization and also minimizes the time features need to be cached while waiting for back-propagation. As a result, storage demand is reduced to the point where only on-chip memory is used for the convolution layers. We evaluate FPDeep with the Alexnet, VGG-16, and VGG-19 benchmarks. Experimental results show that FPDeep has good scalability to a large number of FPGAs, with the limiting factor being the FPGA-to-FPGA bandwidth. With 6 transceivers per FPGA, FPDeep shows linearity up to 83 FPGAs. Energy efficiency is evaluated with respect to GOPs/J. FPDeep provides, on average, 6.36x higher energy efficiency than comparable GPU servers.
Tasks
Published	2019-01-04
URL	http://arxiv.org/abs/1901.01007v1
PDF	http://arxiv.org/pdf/1901.01007v1.pdf
PWC	https://paperswithcode.com/paper/a-scalable-framework-for-acceleration-of-cnn
Repo
Framework

Low-Rank Approximation of Matrices for PMI-based Word Embeddings


Title	Low-Rank Approximation of Matrices for PMI-based Word Embeddings
Authors	Alena Sorokina, Aidana Karipbayeva, Zhenisbek Assylbekov
Abstract	We perform an empirical evaluation of several methods of low-rank approximation in the problem of obtaining PMI-based word embeddings. All word vectors were trained on parts of a large corpus extracted from English Wikipedia (enwik9) which was divided into two equal-sized datasets, from which PMI matrices were obtained. A repeated measures design was used in assigning a method of low-rank approximation (SVD, NMF, QR) and dimensionality of the vectors (250, 500) to each of the PMI matrix replicates. Our experiments show that word vectors obtained from the truncated SVD achieve the best performance on two downstream tasks, similarity and analogy, compare to the other two low-rank approximation methods.
Tasks	Word Embeddings
Published	2019-09-21
URL	https://arxiv.org/abs/1909.09855v1
PDF	https://arxiv.org/pdf/1909.09855v1.pdf
PWC	https://paperswithcode.com/paper/190909855
Repo
Framework

Incremental ELMVIS for unsupervised learning


Title	Incremental ELMVIS for unsupervised learning
Authors	Anton Akusok, Emil Eirola, Yoan Miche, Ian Oliver, Kaj-Mikael Björk, Andrey Gritsenko, Stephen Baek, Amaury Lendasse
Abstract	An incremental version of the ELMVIS+ method is proposed in this paper. It iteratively selects a few best fitting data samples from a large pool, and adds them to the model. The method keeps high speed of ELMVIS+ while allowing for much larger possible sample pools due to lower memory requirements. The extension is useful for reaching a better local optimum with greedy optimization of ELMVIS, and the data structure can be specified in semi-supervised optimization. The major new application of incremental ELMVIS is not to visualization, but to a general dataset processing. The method is capable of learning dependencies from non-organized unsupervised data – either reconstructing a shuffled dataset, or learning dependencies in complex high-dimensional space. The results are interesting and promising, although there is space for improvements.
Tasks
Published	2019-12-18
URL	https://arxiv.org/abs/1912.08638v1
PDF	https://arxiv.org/pdf/1912.08638v1.pdf
PWC	https://paperswithcode.com/paper/incremental-elmvis-for-unsupervised-learning
Repo
Framework

Distance-Based Approaches to Repair Semantics in Ontology-based Data Access


Title	Distance-Based Approaches to Repair Semantics in Ontology-based Data Access
Authors	César Prouté, Bruno Yun, Madalina Croitoru
Abstract	In the presence of inconsistencies, repair techniques thrive to restore consistency by reasoning with several repairs. However, since the number of repairs can be large, standard inconsistent tolerant semantics usually yield few answers. In this paper, we use the notion of syntactic distance between repairs following the intuition that it can allow us to cluster some repairs “close” to each other. In this way, we propose a generic framework to answer queries in a more personalise fashion.
Tasks
Published	2019-10-01
URL	https://arxiv.org/abs/1910.00293v1
PDF	https://arxiv.org/pdf/1910.00293v1.pdf
PWC	https://paperswithcode.com/paper/distance-based-approaches-to-repair-semantics
Repo
Framework

Region of Interest Segmentation from Lidar Point Cloud for Multirotor Aerial Vehicles


Title	Region of Interest Segmentation from Lidar Point Cloud for Multirotor Aerial Vehicles
Authors	Geesara Prathap, Roman Fedorenko, Alexandr Klimchik
Abstract	We propose a novel filter for segmenting the region of interest from lidar 3D point cloud for multirotor aerial vehicles. It is specially targeted for real-time applications and works on sparse lidar point clouds without preliminary mapping. We use this filter as a crucial component of fast obstacle avoidance system for agriculture drone operating at low altitude. As the first step, a point cloud is transformed into a depth image and then places with a high density near to the vehicle (local maxima) are identified. Then we merge original depth image with identified locations after maximizing intensities of pixels in which local maxima were obtained. Next step is to calculate the range angle image that represents angles between two consecutive laser beams based on the improved depth image. Once a range angle image is constructed, smoothing is applied to reduce the noise. Finally, we find out connected components in the improved depth image while incorporating smoothed range angle image. This allows separating the region of interest. The filter has been tested on a simulated environment as well as an actual drone and provides real-time performance. We make our source code and dataset available online\footnote[2]{Source code and dataset are available at \url{https://github.com/GPrathap/hagen.git}}. Real world experiment result can be found \footnote[3]{Real-world experiment result can be found on the following link: \url{https://www.youtube.com/watch?v=iHd_ZkhKPjc}}
Tasks
Published	2019-11-16
URL	https://arxiv.org/abs/1911.06994v2
PDF	https://arxiv.org/pdf/1911.06994v2.pdf
PWC	https://paperswithcode.com/paper/ground-and-non-ground-separation-filter-for
Repo
Framework

Dying Experts: Efficient Algorithms with Optimal Regret Bounds


Title	Dying Experts: Efficient Algorithms with Optimal Regret Bounds
Authors	Hamid Shayestehmanesh, Sajjad Azami, Nishant A. Mehta
Abstract	We study a variant of decision-theoretic online learning in which the set of experts that are available to Learner can shrink over time. This is a restricted version of the well-studied sleeping experts problem, itself a generalization of the fundamental game of prediction with expert advice. Similar to many works in this direction, our benchmark is the ranking regret. Various results suggest that achieving optimal regret in the fully adversarial sleeping experts problem is computationally hard. This motivates our relaxation where any expert that goes to sleep will never again wake up. We call this setting “dying experts” and study it in two different cases: the case where the learner knows the order in which the experts will die and the case where the learner does not. In both cases, we provide matching upper and lower bounds on the ranking regret in the fully adversarial setting. Furthermore, we present new, computationally efficient algorithms that obtain our optimal upper bounds.
Tasks
Published	2019-10-29
URL	https://arxiv.org/abs/1910.13521v1
PDF	https://arxiv.org/pdf/1910.13521v1.pdf
PWC	https://paperswithcode.com/paper/dying-experts-efficient-algorithms-with
Repo
Framework

Random Machines: A bagged-weighted support vector model with free kernel choice


Title	Random Machines: A bagged-weighted support vector model with free kernel choice
Authors	Anderson Ara, Mateus Maia, Samuel Macêdo, Francisco Louzada
Abstract	Improvement of statistical learning models in order to increase efficiency in solving classification or regression problems is still a goal pursued by the scientific community. In this way, the support vector machine model is one of the most successful and powerful algorithms for those tasks. However, its performance depends directly from the choice of the kernel function and their hyperparameters. The traditional choice of them, actually, can be computationally expensive to do the kernel choice and the tuning processes. In this article, it is proposed a novel framework to deal with the kernel function selection called Random Machines. The results improved accuracy and reduced computational time. The data study was performed in simulated data and over 27 real benchmarking datasets.
Tasks
Published	2019-11-21
URL	https://arxiv.org/abs/1911.09411v1
PDF	https://arxiv.org/pdf/1911.09411v1.pdf
PWC	https://paperswithcode.com/paper/random-machines-a-bagged-weighted-support
Repo
Framework