In this work, we learn a shared encoding representation for a multi-task neural network model optimized with connectionist temporal classification (CTC) and conventional framewise cross-entropy training criteria. Our experiments show that the multi-task training not only tackles the complexity of optimizing CTC models such as acoustic-to-word but also results in significant improvement compared to the plain-task training with an optimal setup. Furthermore, we propose to use the encoding representation learned by the multi-task network to initialize the encoder of attention-based models. Thereby, we train a deep attention-based end-to-end model with 10 long short-term memory (LSTM) layers of encoder which produces 12.2\% and 22.6\% word-error-rate on Switchboard and CallHome subsets of the Hub5 2000 evaluation.
https://arxiv.org/abs/1904.02147
3D point-cloud recognition with PointNet and its variants has received remarkable progress. A missing ingredient, however, is the ability to automatically evaluate point-wise importance w.r.t.! classification performance, which is usually reflected by a saliency map. A saliency map is an important tool as it allows one to perform further processes on point-cloud data. In this paper, we propose a novel way of characterizing critical points and segments to build point-cloud saliency maps. Our method assigns each point a score reflecting its contribution to the model-recognition loss. The saliency map explicitly explains which points are the key for model recognition. Furthermore, aggregations of highly-scored points indicate important segments/subsets in a point-cloud. Our motivation for constructing a saliency map is by point dropping, which is a non-differentiable operator. To overcome this issue, we approximate point-dropping with a differentiable procedure of shifting points towards the cloud centroid. Consequently, each saliency score can be efficiently measured by the corresponding gradient of the loss w.r.t the point under the spherical coordinates. Extensive evaluations on several state-of-the-art point-cloud recognition models, including PointNet, PointNet++ and DGCNN, demonstrate the veracity and generality of our proposed saliency map. Code for experiments is released on \url{https://github.com/tianzheng4/PointCloud-Saliency-Maps}.
http://arxiv.org/abs/1812.01687
We explore deep Reinforcement Learning(RL) algorithms for scalping trading and knew that there is no appropriate trading gym and agent examples. Thus we propose gym and agent like Open AI gym in finance. Not only that, we introduce new RL framework based on our hybrid algorithm which leverages between supervised learning and RL algorithm and uses meaningful observations such order book and settlement data from experience watching scalpers trading. That is very crucial information for traders behavior to be decided. To feed these data into our model, we use spatio-temporal convolution layer, called Conv3D for order book data and temporal CNN, called Conv1D for settlement data. Those are preprocessed by episode filter we developed. Agent consists of four sub agents divided to clarify their own goal to make best decision. Also, we adopted value and policy based algorithm to our framework. With these features, we could make agent mimic scalpers as much as possible. In many fields, RL algorithm has already begun to transcend human capabilities in many domains. This approach could be a starting point to beat human in the financial stock market, too and be a good reference for anyone who wants to design RL algorithm in real world domain. Finally, weexperiment our framework and gave you experiment progress.
http://arxiv.org/abs/1904.00441
Automatic methods for generating state-of-the-art neural network architectures without human experts have generated significant attention recently. This is because of the potential to remove human experts from the design loop which can reduce costs and decrease time to model deployment. Neural architecture search (NAS) techniques have improved significantly in their computational efficiency since the original NAS was proposed. This reduction in computation is enabled via weight sharing such as in Efficient Neural Architecture Search (ENAS). However, recently a body of work confirms our discovery that ENAS does not do significantly better than random search with weight sharing, contradicting the initial claims of the authors. We provide an explanation for this phenomenon by investigating the interpretability of the ENAS controller’s hidden state. We are interested in seeing if the controller embeddings are predictive of any properties of the final architecture - for example, graph properties like the number of connections, or validation performance. We find models sampled from identical controller hidden states have no correlation in various graph similarity metrics. This failure mode implies the RNN controller does not condition on past architecture choices. Importantly, we may need to condition on past choices if certain connection patterns prevent vanishing or exploding gradients. Lastly, we propose a solution to this failure mode by forcing the controller’s hidden state to encode pasts decisions by training it with a memory buffer of previously sampled architectures. Doing this improves hidden state interpretability by increasing the correlation controller hidden states and graph similarity metrics.
https://arxiv.org/abs/1904.00438
In contexts such as teleoperation, robot reprogramming, human-robot-interaction, and neural prosthetics, conveying movement commands to a robotic platform is often a limiting factor. Currently, many applications rely on joint-angle-by-joint-angle prescriptions. This inherently requires a large number of parameters to be specified by the user that scales with the number of degrees of freedom on a platform, creating high bandwidth requirements for interfaces. This paper presents an efficient representation of high-level, spatial commands that specifies many joint angles with relatively few parameters based on a spatial architecture that is judged favorably by human viewers. In particular, a general method for labeling connected platform linkages, generating a databank of user-specified poses, and mapping between high-level spatial commands and specific platform static configurations are presented. Thus, this architecture is ``platform-invariant’’ where the same high-level, spatial command can be executed on any platform. This has the advantage that our commands have meaning for human movers as well. In order to achieve this, we draw inspiration from Laban/Bartenieff Movement Studies, an embodied taxonomy for movement description. The architecture is demonstrated through implementation on 26 spatial directions for a Rethink Robotics Baxter, an Aldebaran NAO, and a KUKA youBot. User studies are conducted to validate the claims of the proposed framework.
http://arxiv.org/abs/1904.00424
One-shot method is a powerful Neural Architecture Search (NAS) framework, but its training is non-trivial and it is difficult to achieve competitive results on large scale datasets like ImageNet. In this work, we propose a Single Path One-Shot model to address its main challenge in the training. Our central idea is to construct a simplified supernet, Single Path Supernet, which is trained by an uniform path sampling method. All underlying architectures (and their weights) get trained fully and equally. Once we have a trained supernet, we apply an evolutionary algorithm to efficiently search the best-performing architectures without any fine tuning. Comprehensive experiments verify that our approach is flexible and effective. It is easy to train and fast to search. It effortlessly supports complex search spaces (e.g., building blocks, channel, mixed-precision quantization) and different search constraints (e.g., FLOPs, latency). It is thus convenient to use for various needs. It achieves start-of-the-art performance on the large dataset ImageNet.
http://arxiv.org/abs/1904.00420
Occupancy grid mapping is an important component of autonomous vehicle perception. It encapsulates information of the drivable area, road obstacles and enables safe autonomous driving. To this end, radars are becoming widely used due to their long range sensing, low cost, and robustness to severe weather conditions. Despite recent advances in deep learning technology, occupancy grid mapping from radar data is still mostly done using classical filtering approaches. In this work, we propose a data driven approach for learning an inverse sensor model used for occupancy grid mapping from clustered radar data. This task is very challenging due to data sparsity and noise characteristics of the radar sensor. The problem is formulated as a semantic segmentation task and we show how it can be learned in a self-supervised manner using lidar data for generating ground truth. We show both qualitatively and quantitatively that our learned occupancy net outperforms classic methods by a large margin using the recently released NuScenes real-world driving data.
http://arxiv.org/abs/1904.00415
We introduce the first work to tackle the image retrieval problem as a continuous operation. While the proposed approaches in the literature can be roughly categorized into two main groups: category- and instance-based retrieval, in this work we show that the retrieval task is much richer and more complex. Image similarity goes beyond this discrete vantage point and spans a continuous spectrum among the classical operating points of category and instance similarity. However, current retrieval models are static and incapable of exploring this rich structure of the retrieval space since they are trained and evaluated with a single operating point as a target objective. Hence, we introduce a novel retrieval model that for a given query is capable of producing a dynamic embedding that can target an arbitrary point along the continuous retrieval spectrum. Our model disentangles the visual signal of a query image into its basic components of categorical and attribute information. Furthermore, using a continuous control parameter our model learns to reconstruct a dynamic embedding of the query by mixing these components with different proportions to target a specific point along the retrieval simplex. We demonstrate our idea in a comprehensive evaluation of the proposed model and highlight the advantages of our approach against a set of well-established discrete retrieval models.
http://arxiv.org/abs/1812.00202
To solve the red jujube classification problem, this paper designs a convolutional neural network model with low computational cost and high classification accuracy. The architecture of the model is inspired by the multi-visual mechanism of the organism and DenseNet. To further improve our model, we add the attention mechanism of SE-Net. We also construct a dataset which contains 23,735 red jujube images captured by a jujube grading system. According to the appearance of the jujube and the characteristics of the grading system, the dataset is divided into four classes: invalid, rotten, wizened and normal. The numerical experiments show that the classification accuracy of our model reaches to 91.89%, which is comparable to DenseNet-121, InceptionV3, InceptionV4, and Inception-ResNet v2. However, our model has real-time performance.
http://arxiv.org/abs/1904.00388
With the rapid development of deep convolutional neural network, face detection has made great progress in recent years. WIDER FACE dataset, as a main benchmark, contributes greatly to this area. A large amount of methods have been put forward where PyramidBox designs an effective data augmentation strategy (Data-anchor-sampling) and context-based module for face detector. In this report, we improve each part to further boost the performance, including Balanced-data-anchor-sampling, Dual-PyramidAnchors and Dense Context Module. Specifically, Balanced-data-anchor-sampling obtains more uniform sampling of faces with different sizes. Dual-PyramidAnchors facilitate feature learning by introducing progressive anchor loss. Dense Context Module with dense connection not only enlarges receptive filed, but also passes information efficiently. Integrating these techniques, PyramidBox++ is constructed and achieves state-of-the-art performance in hard set.
http://arxiv.org/abs/1904.00386
The main motivation of this work is to propose a simulation approach for a specific task within the UAV (Unmanned Aerial Vehicle) field, i.e., the visual detection and tracking of arbitrary moving objects. In particular, it is described MAT-Fly, a numerical simulation platform for multi-rotors aircrafts characterized by ease of use and control development. The platform is based on Matlab and the MathWorks Virtual Reality (VR) and Computer Vision System (CVS) toolboxes that work together to simulate the behavior of a drone in a 3D environment while tracking a car that moves along a non trivial path. The VR toolbox has been chosen due to the familiarity that students have with Matlab and because it allows to move the attention to the classifier, the tracker, the reference generator and the trajectory tracking control thanks to its simple structure. The overall architecture is quite modular so that each block can be easily replaced with others by simplifying the development phase and by allowing to add even more functionalities. The numerical simulator makes easy and quick to insert and to remove flight control system components, testing and comparing different plans, both for indoor and outdoor scenarios, when computer vision algorithms are in the loop. In an automatic way, the simulator is able to acquire frames from the virtual world, to search for one or more objects on which it has been trained during the learning phase, and to track the target position applying a trajectory control, addressing in that way the image-based visual servoing problems. Some simple testbeds have been presented in order to show the effectiveness and robustness of the approach as well as the platform works. We released the software as open-source, making it available for educational activities.
http://arxiv.org/abs/1904.00378
Active learning aims to develop label-efficient algorithms by sampling the most representative queries to be labeled by an oracle. We describe a pool-based semi-supervised active learning algorithm that implicitly learns this sampling mechanism in an adversarial manner. Our method learns a latent space using a variational autoencoder (VAE) and an adversarial network trained to discriminate between unlabeled and labeled data. The mini-max game between the VAE and the adversarial network is played such that while the VAE tries to trick the adversarial network into predicting that all data points are from the labeled pool, the adversarial network learns how to discriminate between dissimilarities in the latent space. We extensively evaluate our method on various image classification and semantic segmentation benchmark datasets and establish a new state of the art on $\text{CIFAR10/100}$, $\text{Caltech-256}$, $\text{ImageNet}$, $\text{Cityscapes}$, and $\text{BDD100K}$. Our results demonstrate that our adversarial approach learns an effective low dimensional latent space in large-scale settings and provides for a computationally efficient sampling method.
http://arxiv.org/abs/1904.00370
There is a huge imbalance between languages currently spoken and corresponding resources to study them. Most of the attention naturally goes to the “big” languages: those which have the largest presence in terms of media and number of speakers. Other less represented languages sometimes do not even have a good quality corpus to study them. In this paper, we tackle this imbalance by presenting a new set of evaluation resources for Tatar, a language of the Turkic language family which is mainly spoken in Tatarstan Republic, Russia. We present three datasets: Similarity and Relatedness datasets that consist of human scored word pairs and can be used to evaluate semantic models; and Analogies dataset that comprises analogy questions and allows to explore semantic, syntactic, and morphological aspects of language modeling. All three datasets build upon existing datasets for the English language and follow the same structure. However, they are not mere translations. They take into account specifics of the Tatar language and expand beyond the original datasets. We evaluate state-of-the-art word embedding models for two languages using our proposed datasets for Tatar and the original datasets for English and report our findings on performance comparison.
http://arxiv.org/abs/1904.00365
Deep part-based methods in recent literature have revealed the great potential of learning local part-level representation for pedestrian image in the task of person re-identification. However, global features that capture discriminative holistic information of human body are usually ignored or not well exploited. This motivates us to investigate joint learning global and local features from pedestrian images. Specifically, in this work, we propose a novel framework termed tree branch network (TBN) for person re-identification. Given a pedestrain image, the feature maps generated by the backbone CNN, are partitioned recursively into several pieces, each of which is followed by a bottleneck structure that learns finer-grained features for each level in the hierarchical tree-like framework. In this way, representations are learned in a coarse-to-fine manner and finally assembled to produce more discriminative image descriptions. Experimental results demonstrate the effectiveness of the global and local feature learning method in the proposed TBN framework. We also show significant improvement in performance over state-of-the-art methods on three public benchmarks: Market-1501, CUHK-03 and DukeMTMC.
http://arxiv.org/abs/1904.00355
Restoring a sharp light field image from its blurry input has become essential due to the increasing popularity of parallax-based image processing. State-of-the-art blind light field deblurring methods suffer from several issues such as slow processing, reduced spatial size, and a limited motion blur model. In this work, we address these challenging problems by generating a complex blurry light field dataset and proposing a learning-based deblurring approach. In particular, we model the full 6-degree of freedom (6-DOF) light field camera motion, which is used to create the blurry dataset using a combination of real light fields captured with a Lytro Illum camera, and synthetic light field renderings of 3D scenes. Furthermore, we propose a light field deblurring network that is built with the capability of large receptive fields. We also introduce a simple strategy of angular sampling to train on the large-scale blurry light field effectively. We evaluate our method through both quantitative and qualitative measurements and demonstrate superior performance compared to the state-of-the-art method with a massive speedup in execution time. Our method is about 16K times faster than Srinivasan et. al. [22] and can deblur a full-resolution light field in less than 2 seconds.
http://arxiv.org/abs/1904.00352
The recent surge of text-based online counseling applications enables us to collect and analyze interactions between counselors and clients. A dataset of those interactions can be used to learn to automatically classify the client utterances into categories that help counselors in diagnosing client status and predicting counseling outcome. With proper anonymization, we collect counselor-client dialogues, define meaningful categories of client utterances with professional counselors, and develop a novel neural network model for classifying the client utterances. The central idea of our model, ConvMFiT, is a pre-trained conversation model which consists of a general language model built from an out-of-domain corpus and two role-specific language models built from unlabeled in-domain dialogues. The classification result shows that ConvMFiT outperforms state-of-the-art comparison models. Further, the attention weights in the learned model confirm that the model finds expected linguistic patterns for each category.
http://arxiv.org/abs/1904.00350
Benefitted from its great success on many tasks, deep learning is increasingly used on low-computational-cost devices, e.g. smartphone, embedded devices, etc. To reduce the high computational and memory cost, in this work, we propose a fully learnable group convolution module (FLGC for short) which is quite efficient and can be embedded into any deep neural networks for acceleration. Specifically, our proposed method automatically learns the group structure in the training stage in a fully end-to-end manner, leading to a better structure than the existing pre-defined, two-steps, or iterative strategies. Moreover, our method can be further combined with depthwise separable convolution, resulting in 5 times acceleration than the vanilla Resnet50 on single CPU. An additional advantage is that in our FLGC the number of groups can be set as any value, but not necessarily 2^k as in most existing methods, meaning better tradeoff between accuracy and speed. As evaluated in our experiments, our method achieves better performance than existing learnable group convolution and standard group convolution when using the same number of groups.
http://arxiv.org/abs/1904.00346
In this paper, we propose a new model to segment cells in phase contrast microscopy images. Cell images collected from the similar scenario share a similar background. Inspired by this, we separate cells from the background in images by formulating the problem as a low-rank and structured sparse matrix decomposition problem. Then, we propose the inverse diffraction pattern filtering method to further segment individual cells in the images. This is a deconvolution process that has a much lower computational complexity when compared to the other restoration methods. Experiments demonstrate the effectiveness of the proposed model when it is compared with recent works.
http://arxiv.org/abs/1904.00328
Laboratory testing and medication prescription are two of the most important routines in daily clinical practice. Developing an artificial intelligence system that can automatically make lab test imputations and medication recommendations can save cost on potentially redundant lab tests and inform physicians in more effective prescription. We present an intelligent model that can automatically recommend the patients’ medications based on their incomplete lab tests, and can even accurately estimate the lab values that have not been taken. We model the complex relations between multiple types of medical entities with their inherent features in a heterogeneous graph. Then we learn a distributed representation for each entity in the graph based on graph convolutional networks to make the representations integrate information from multiple types of entities. Since the entity representations incorporate multiple types of medical information, they can be used for multiple medical tasks. In our experiments, we construct a graph to associate patients, encounters, lab tests and medications, and conduct the two tasks: medication recommendation and lab test imputation. The experimental results demonstrate that our model can outperform the state-of-the-art models in both tasks.
http://arxiv.org/abs/1904.00326
Image representation is a fundamental task in computer vision. However, most of the existing approaches for image representation ignore the relations between images and consider each input image independently. Intuitively, relations between images can help to understand the images and maintain model consistency over related images. In this paper, we consider modeling the image-level relations to generate more informative image representations, and propose ImageGCN, an end-to-end graph convolutional network framework for multi-relational image modeling. We also apply ImageGCN to chest X-ray (CXR) images where rich relational information is available for disease identification. Unlike previous image representation models, ImageGCN learns the representation of an image using both its original pixel features and the features of related images. Besides learning informative representations for images, ImageGCN can also be used for object detection in a weakly supervised manner. The Experimental results on ChestX-ray14 dataset demonstrate that ImageGCN can outperform respective baselines in both disease identification and localization tasks and can achieve comparable and often better results than the state-of-the-art methods.
http://arxiv.org/abs/1904.00325
Feature correspondence selection is pivotal to many feature-matching based tasks in computer vision. Searching for spatially k-nearest neighbors is a common strategy for extracting local information in many previous works. However, there is no guarantee that the spatially k-nearest neighbors of correspondences are consistent because the spatial distribution of false correspondences is often irregular. To address this issue, we present a compatibility-specific mining method to search for consistent neighbors. Moreover, in order to extract and aggregate more reliable features from neighbors, we propose a hierarchical network named NM-Net with a series of convolution layers taking the generated graph as input, which is insensitive to the order of correspondences. Our experimental results have shown the proposed method achieves the state-of-the-art performance on four datasets with various inlier ratios and varying numbers of feature consistencies.
http://arxiv.org/abs/1904.00320
Despite the recent active research on processing point clouds with deep networks, few attention has been on the sensitivity of the networks to rotations. In this paper, we propose a deep learning architecture that achieves discrete $\mathbf{SO}(2)$/$\mathbf{SO}(3)$ rotation equivariance for point cloud recognition. Specifically, the rotation of an input point cloud with elements of a rotation group is similar to shuffling the feature vectors generated by our approach. The equivariance is easily reduced to invariance by eliminating the permutation with operations such as maximum or average. Our method can be directly applied to any existing point cloud based networks, resulting in significant improvements in their performance for rotated inputs. We show state-of-the-art results in the classification tasks with various datasets under both $\mathbf{SO}(2)$ and $\mathbf{SO}(3)$ rotations. In addition, we further analyze the necessary conditions of applying our approach to PointNet based networks. Source codes at https://github.com/lijx10/rot-equ-net
http://arxiv.org/abs/1904.00319
When ontologies reach a certain size and complexity, faults such as inconsistencies, unsatisfiable classes or wrong entailments are hardly avoidable. Locating the incorrect axioms that cause these faults is a hard and time-consuming task. Addressing this issue, several techniques for semi-automatic fault localization in ontologies have been proposed. Often, these approaches involve a human expert who provides answers to system-generated questions about the intended (correct) ontology in order to reduce the possible fault locations. To suggest as informative questions as possible, existing methods draw on various algorithmic optimizations as well as heuristics. However, these computations are often based on certain assumptions about the interacting user. In this work, we characterize and discuss different user types and show that existing approaches do not achieve optimal efficiency for all of them. As a remedy, we suggest a new type of expert question which aims at fitting the answering behavior of all analyzed experts. Moreover, we present an algorithm to optimize this new query type which is fully compatible with the (tried and tested) heuristics used in the field. Experiments on faulty real-world ontologies show the potential of the new querying method for minimizing the expert consultation time, independent of the expert type. Besides, the gained insights can inform the design of interactive debugging tools towards better meeting their users’ needs.
http://arxiv.org/abs/1904.00317
FDA drug labels are rich sources of information about drugs and drug-disease relations, but their complexity makes them challenging texts to analyze in isolation. To overcome this, we situate these labels in two health knowledge graphs: one built from precise structured information about drugs and diseases, and another built entirely from a database of clinical narrative texts using simple heuristic methods. We show that Probabilistic Soft Logic models defined over these graphs are superior to text-only and relation-only variants, and that the clinical narratives graph delivers exceptional results with little manual effort. Finally, we release a new dataset of drug labels with annotations for five distinct drug-disease relations.
http://arxiv.org/abs/1904.00313
Addressing catastrophic forgetting is one of the key challenges in continual learning where machine learning systems are trained with sequential or streaming tasks. Despite recent remarkable progress in state-of-the-art deep learning, deep neural networks (DNNs) are still plagued with the catastrophic forgetting problem. This paper presents a conceptually simple yet general and effective framework for handling catastrophic forgetting in continual learning with DNNs. The proposed method consists of two components: a neural structure optimization component and a parameter learning and/or fine-tuning component. The former learns the best neural structure for the current task on top of the current DNN trained with previous tasks. It learns whether to reuse or adapt building blocks in the current DNN, or to create new ones if needed under the differentiable neural architecture search framework. The latter estimates parameters for newly introduced structures, and fine-tunes the old ones if preferred. By separating the explicit neural structure learning and the parameter estimation, not only is the proposed method capable of evolving neural structures in an intuitively meaningful way, but also shows strong capabilities of alleviating catastrophic forgetting in experiments. Furthermore, the proposed method outperforms all other baselines on the permuted MNIST dataset, the split CIFAR100 dataset and the Visual Domain Decathlon dataset in continual learning setting.
http://arxiv.org/abs/1904.00310
In this work, we present an interaction-based approach to learn semantically rich representations for the task of slicing vegetables. Unlike previous approaches, we focus on object-centric representations and use auxiliary tasks to learn rich representations using a two-step process. First, we use simple auxiliary tasks, such as predicting the thickness of a cut slice, to learn an embedding space which captures object properties that are important for the task of slicing vegetables. In the second step, we use these learned latent embeddings to learn a forward model. Learning a forward model affords us to plan online in the latent embedding space and forces our model to improve its representations while performing the slicing task. To show the efficacy of our approach we perform experiments on two different vegetables: cucumbers and tomatoes. Our experimental evaluation shows that our method is able to capture important semantic properties for the slicing task, such as the thickness of the vegetable being cut. We further show that by using our learned forward model, we can plan for the task of vegetable slicing.
http://arxiv.org/abs/1904.00303
This paper presents a educational workshop in Scratch that is proposed for the active participation of undergraduate students in contexts of Artificial Intelligence. The main objective of the activity is to demystify the complexity of Artificial Intelligence and its algorithms. For this purpose, students must realize simple exercises of clustering and two neural networks, in Scratch. The detailed methodology to get that is presented in the article.
http://arxiv.org/abs/1904.00296
Long short-term memory (LSTM) and recurrent neural network (RNN) has achieved great successes on time-series prediction. In this paper, a methodology of using LSTM-based deep-RNN for two-phase flow regime prediction is proposed, motivated by previous research on constructing deep RNN. The method is featured with fast response and accuracy. The built RNN networks are trained and tested with time-series void fraction data collected using impedance void meter. The result shows that the prediction accuracy depends on the depth of network and the number of layer cells. However, deeper and larger network consumes more time in predicting.
http://arxiv.org/abs/1904.00291
Deep convolutional neural networks (CNNs) are widely known for their outstanding performance in classification and regression tasks over high-dimensional data. This made them a popular and powerful tool for a large variety of applications in industry and academia. Recent publications show that seemingly easy classifaction tasks (for humans) can be very challenging for state of the art CNNs. An attempt to describe how humans perceive visual elements is given by the Gestalt principles. In this paper we evaluate AlexNet and GoogLeNet regarding their performance on classifying the correctness of the well known Kanizsa triangles, which heavily rely on the Gestalt principle of closure. Therefore we created various datasets containing valid as well as invalid variants of the Kanizsa triangle. Our findings suggest that perceiving objects by utilizing the principle of closure is very challenging for the applied network architectures but they appear to adapt to the effect of closure.
http://arxiv.org/abs/1904.00285
We propose a stochastic answer network (SAN) to explore multi-step inference strategies in Natural Language Inference. Rather than directly predicting the results given the inputs, the model maintains a state and iteratively refines its predictions. Our experiments show that SAN achieves the state-of-the-art results on three benchmarks: Stanford Natural Language Inference (SNLI) dataset, MultiGenre Natural Language Inference (MultiNLI) dataset and Quora Question Pairs dataset.
http://arxiv.org/abs/1804.07888
Humans can only interact with part of the surrounding environment due to biological restrictions. Therefore, we learn to reason the spatial relationships across a series of observations to piece together the surrounding environment. Inspired by such behavior and the fact that machines also have computational constraints, we propose \underline{CO}nditional \underline{CO}ordinate GAN (COCO-GAN) of which the generator generates images by parts based on their spatial coordinates as the condition. On the other hand, the discriminator learns to justify realism across multiple assembled patches by global coherence, local appearance, and edge-crossing continuity. Despite the full images are never generated during training, we show that COCO-GAN can produce \textbf{state-of-the-art-quality} full images during inference. We further demonstrate a variety of novel applications enabled by teaching the network to be aware of coordinates. First, we perform extrapolation to the learned coordinate manifold and generate off-the-boundary patches. Combining with the originally generated full image, COCO-GAN can produce images that are larger than training samples, which we called “beyond-boundary generation”. We then showcase panorama generation within a cylindrical coordinate system that inherently preserves horizontally cyclic topology. On the computation side, COCO-GAN has a built-in divide-and-conquer paradigm that reduces memory requisition during training and inference, provides high-parallelism, and can generate parts of images on-demand.
http://arxiv.org/abs/1904.00284
WiFi human sensing has achieved great progress in indoor localization, activity classification, etc. Retracing the development of these work, we have a natural question: can WiFi devices work like cameras for vision applications? In this paper We try to answer this question by exploring the ability of WiFi on estimating single person pose. We use a 3-antenna WiFi sender and a 3-antenna receiver to generate WiFi data. Meanwhile, we use a synchronized camera to capture person videos for corresponding keypoint annotations. We further propose a fully convolutional network (FCN), termed WiSPPN, to estimate single person pose from the collected data and annotations. Evaluation on over 80k images (16 sites and 8 persons) replies aforesaid question with a positive answer. Codes have been made publicly available at https://github.com/geekfeiw/WiSPPN.
http://arxiv.org/abs/1904.00277
Fine-grained person perception such as body segmentation and pose estimation has been achieved with many 2D and 3D sensors such as RGB/depth cameras, radars (e.g., RF-Pose) and LiDARs. These sensors capture 2D pixels or 3D point clouds of person bodies with high spatial resolution, such that the existing Convolutional Neural Networks can be directly applied for perception. In this paper, we take one step forward to show that fine-grained person perception is possible even with 1D sensors: WiFi antennas. To our knowledge, this is the first work to perceive persons with pervasive WiFi devices, which is cheaper and power efficient than radars and LiDARs, invariant to illumination, and has little privacy concern comparing to cameras. We used two sets of off-the-shelf WiFi antennas to acquire signals, i.e., one transmitter set and one receiver set. Each set contains three antennas lined-up as a regular household WiFi router. The WiFi signal generated by a transmitter antenna, penetrates through and reflects on human bodies, furniture and walls, and then superposes at a receiver antenna as a 1D signal sample (instead of 2D pixels or 3D point clouds). We developed a deep learning approach that uses annotations on 2D images, takes the received 1D WiFi signals as inputs, and performs body segmentation and pose estimation in an end-to-end manner. Experimental results on over 100000 frames under 16 indoor scenes demonstrate that Person-in-WiFi achieved person perception comparable to approaches using 2D images.
http://arxiv.org/abs/1904.00276
We introduce Tempered Geodesic Markov Chain Monte Carlo (TG-MCMC) algorithm for initializing pose graph optimization problems, arising in various scenarios such as SFM (structure from motion) or SLAM (simultaneous localization and mapping). TG-MCMC is first of its kind as it unites asymptotically global non-convex optimization on the spherical manifold of quaternions with posterior sampling, in order to provide both reliable initial poses and uncertainty estimates that are informative about the quality of individual solutions. We devise rigorous theoretical convergence guarantees for our method and extensively evaluate it on synthetic and real benchmark datasets. Besides its elegance in formulation and theory, we show that our method is robust to missing data, noise and the estimated uncertainties capture intuitive properties of the data.
http://arxiv.org/abs/1805.12279
Hopfions are an intriguing class of string-like solitons, named according to a classical topological concept classifying three-dimensional direction fields. The search of hopfions in real physical systems is going on for nearly half a century, starting with the seminal work of Faddeev. But so far realizations in solids are missing. Here, we present a theory that identifies magnetic materials featuring hopfions as stable states without the assistance of confinement or external fields. Our results are based on an advanced micromagnetic energy functional derived from a spin-lattice Hamiltonian. Hopfions appear as emergent particles of the classical Heisenberg model. Magnetic hopfions represent three-dimensional particle-like objects of nanometre-size dimensions opening the gate to a new generation of spintronic devices in the framework of a truly three-dimensional architecture. Our approach goes beyond the conventional phenomenological models. We derive material-realistic parameters that serve as concrete guidance in the search of magnetic hopfions bridging computational physics with materials science.
https://arxiv.org/abs/1904.00250
In this paper, we propose an online learning approach that enables the inverse dynamics model learned for a source robot to be transferred to a target robot (e.g., from one quadrotor to another quadrotor with different mass or aerodynamic properties). The goal is to leverage knowledge from the source robot such that the target robot achieves high-accuracy trajectory tracking on arbitrary trajectories from the first attempt with minimal data recollection and training. Most existing approaches for multi-robot knowledge transfer are based on post-analysis of datasets collected from both robots. In this work, we study the feasibility of impromptu transfer of models across robots by learning an error prediction module online. In particular, we analytically derive the form of the mapping to be learned by the online module for exact tracking, propose an approach for characterizing similarity between robots, and use these results to analyze the stability of the overall system. The proposed approach is illustrated in simulation and verified experimentally on two different quadrotors performing impromptu trajectory tracking tasks, where the quadrotors are required to accurately track arbitrary hand-drawn trajectories from the first attempt.
http://arxiv.org/abs/1904.00249
Due to the great growth of motorcycles in the urban fleet and the growth of the study on its behavior and of how this vehicle affects the flow of traffic becomes necessary the development of tools and techniques different from the conventional ones to identify its presence in the traffic flow and be able to extract your information. The article in question attempts to contribute to the study on this type of vehicle by generating a motorcycle image bank and developing and calibrating a motorcycle classifier by combining the LBP techniques to create the characteristic vectors and the classification technique LinearSVC to perform the predictions. In this way the classifier of vehicles of the type motorcycle developed in this research can classify the images of vehicles extracted of videos of monitoring between two classes motorcycles and non-motorcycles with a precision and an accuracy superior to 0,9.
http://arxiv.org/abs/1904.00247
Inspired by the effectiveness of adversarial training in the area of Generative Adversarial Networks we present a new approach for learning feature representations in person re-identification. We investigate different types of bias that typically occur in re-ID scenarios, i.e., pose, body part and camera view, and propose a general approach to address them. We introduce an adversarial strategy for controlling bias, named Bias-controlled Adversarial framework (BCA), with two complementary branches to reduce or to enhance bias-related features. The results and comparison to the state of the art on different benchmarks show that our framework is an effective strategy for person re-identification. The performance improvements are in both full and partial views of persons.
http://arxiv.org/abs/1904.00244
It remains challenging to automatically segment kidneys in clinical ultrasound (US) images due to the kidneys’ varied shapes and image intensity distributions, although semi-automatic methods have achieved promising performance. In this study, we propose subsequent boundary distance regression and pixel classification networks to segment the kidneys, informed by the fact that the kidney boundaries have relatively homogenous texture patterns across images. Particularly, we first use deep neural networks pre-trained for classification of natural images to extract high-level image features from US images, then these features are used as input to learn kidney boundary distance maps using a boundary distance regression network, and finally the predicted boundary distance maps are classified as kidney pixels or non-kidney pixels using a pixel classification network in an end-to-end learning fashion. We also adopted a data-augmentation method based on kidney shape registration to generate enriched training data from a small number of US images with manually segmented kidney labels. Experimental results have demonstrated that our method could effectively improve the performance of automatic kidney segmentation, significantly better than deep learning-based pixel classification networks.
http://arxiv.org/abs/1811.04815
We present a visual symptom checker that combines a pre-trained Convolutional Neural Network (CNN) with a Reinforcement Learning (RL) agent as a Question Answering (QA) model. This method increases the classification confidence and accuracy of the visual symptom checker, and decreases the average number of questions asked to narrow down the differential diagnosis. A Deep Q-Network (DQN)-based RL agent learns how to ask the patient about the presence of symptoms in order to maximize the probability of correctly identifying the underlying condition. The RL agent uses the visual information provided by CNN in addition to the answers to the asked questions to guide the QA system. We demonstrate that the RL-based approach increases the accuracy more than 20% compared to the CNN-only approach, which only uses the visual information to predict the condition. Moreover, the increased accuracy is up to 10% compared to the approach that uses the visual information provided by CNN along with a conventional decision tree-based QA system. We finally show that the RL-based approach not only outperforms the decision tree-based approach, but also narrows down the diagnosis faster in terms of the average number of asked questions.
http://arxiv.org/abs/1903.03495
Online signature verification (OSV) is one of the most challenging tasks in writer identification and digital forensics. Owing to the large intra-individual variability, there is a critical requirement to accurately learn the intra-personal variations of the signature to achieve higher classification accuracy. To achieve this, in this paper, we propose an OSV framework based on deep convolutional Siamese network (DCSN). DCSN automatically extracts robust feature descriptions based on metric-based loss function which decreases intra-writer variability (Genuine-Genuine) and increases inter-individual variability (Genuine-Forgery) and directs the DCSN for effective discriminative representation learning for online signatures and extend it for one shot learning framework. Comprehensive experimentation conducted on three widely accepted benchmark datasets MCYT-100 (DB1), MCYT-330 (DB2) and SVC-2004-Task2 demonstrate the capability of our framework to distinguish the genuine and forgery samples. Experimental results confirm the efficiency of deep convolutional Siamese network based OSV by achieving a lower error rate as compared to many recent and state-of-the art OSV techniques.
http://arxiv.org/abs/1904.00240
We explore active learning (AL) for improving the accuracy of new domains in a natural language understanding (NLU) system. We propose an algorithm called Majority-CRF that uses an ensemble of classification models to guide the selection of relevant utterances, as well as a sequence labeling model to help prioritize informative examples. Experiments with three domains show that Majority-CRF achieves 6.6%-9% relative error rate reduction compared to random sampling with the same annotation budget, and statistically significant improvements compared to other AL approaches. Additionally, case studies with human-in-the-loop AL on six new domains show 4.6%-9% improvement on an existing NLU system.
http://arxiv.org/abs/1810.03450
Autonomous driving decision-making is a great challenge due to the complexity and uncertainty of the traffic environment. Combined with the rule-based constraints, a Deep Q-Network (DQN) based method is applied for autonomous driving lane change decision-making task in this study. Through the combination of high-level lateral decision-making and low-level rule-based trajectory modification, a safe and efficient lane change behavior can be achieved. With the setting of our state representation and reward function, the trained agent is able to take appropriate actions in a real-world-like simulator. The generated policy is evaluated on the simulator for 10 times, and the results demonstrate that the proposed rule-based DQN method outperforms the rule-based approach and the DQN method.
http://arxiv.org/abs/1904.00231
We present a self-supervised task on point clouds, in order to learn meaningful point-wise features that encode local structure around each point. Our self-supervised network, named MortonNet, operates directly on unstructured/unordered point clouds. Using a multi-layer RNN, MortonNet predicts the next point in a point sequence created by a popular and fast Space Filling Curve, the Morton-order curve. The final RNN state (coined Morton feature) is versatile and can be used in generic 3D tasks on point clouds. In fact, we show how Morton features can be used to significantly improve performance (+3% for 2 popular semantic segmentation algorithms) in the task of semantic segmentation of point clouds on the challenging and large-scale S3DIS dataset. We also show how MortonNet trained on S3DIS transfers well to another large-scale dataset, vKITTI, leading to an improvement over state-of-the-art of 3.8%. Finally, we use Morton features to train a much simpler and more stable model for part segmentation in ShapeNet. Our results show how our self-supervised task results in features that are useful for 3D segmentation tasks, and generalize well to other datasets.
http://arxiv.org/abs/1904.00230
In this paper, we propose the USIP detector: an Unsupervised Stable Interest Point detector that can detect highly repeatable and accurately localized keypoints from 3D point clouds under arbitrary transformations without the need for any ground truth training data. Our USIP detector consists of a feature proposal network that learns stable keypoints from input 3D point clouds and their respective transformed pairs from randomly generated transformations. We provide degeneracy analysis of our USIP detector and suggest solutions to prevent it. We encourage high repeatability and accurate localization of the keypoints with a probabilistic chamfer loss that minimizes the distances between the detected keypoints from the training point cloud pairs. Extensive experimental results of repeatability tests on several simulated and real-world 3D point cloud datasets from Lidar, RGB-D and CAD models show that our USIP detector significantly outperforms existing hand-crafted and deep learning-based 3D keypoint detectors. Our code is available at the project website. https://github.com/lijx10/USIP
http://arxiv.org/abs/1904.00229
Video action detectors are usually trained using video datasets with fully supervised temporal annotations. Building such video datasets is a heavily expensive task. To alleviate this problem, recent algorithms leverage weak labelling where videos are untrimmed and only a video-level label is available. In this paper, we propose RefineLoc, a new method for weakly-supervised temporal action localization. RefineLoc uses an iterative refinement approach by estimating and training on snippet-level pseudo ground truth at every iteration. We show the benefit of using such an iterative approach and present an extensive analysis of different pseudo ground truth generators. We show the effectiveness of our model on two standard action datasets, ActivityNet v1.2 and THUMOS14. RefineLoc equipped with a segment prediction-based pseudo ground truth generator improves the state-of-the-art in weakly-supervised temporal localization on the challenging and large-scale ActivityNet dataset by 4.2% and achieves comparable performance with state-of-the-art on THUMOS14.
http://arxiv.org/abs/1904.00227
Convolutional neural network (CNN) architectures have traditionally been explored by human experts in a manual search process that is time-consuming and ineffectively explores the massive space of potential solutions. Neural architecture search (NAS) methods automatically search the space of neural network hyperparameters in order to find optimal task-specific architectures. NAS methods have discovered CNN architectures that achieve state-of-the-art performance in image classification among other tasks, however the application of NAS to image-to-image regression problems such as image restoration is sparse. This paper proposes a NAS method that performs computationally efficient evolutionary search of a minimally constrained network architecture search space. The performance of architectures discovered by the proposed method is evaluated on a variety of image restoration tasks applied to the ImageNet64x64 dataset, and compared with human-engineered CNN architectures. The best neural architectures discovered using only 2 GPU-hours of evolutionary search exhibit comparable performance to the human-engineered baseline architecture.
https://arxiv.org/abs/1812.05866
Deep Convolutional Neural Network (CNN) features have been demonstrated to be effective perceptual quality features. The perceptual loss, based on feature maps of pre-trained CNN’s has proven to be remarkably effective for CNN based perceptual image restoration problems. In this work, taking inspiration from the the Human Visual System (HVS) and our visual perception, we propose a spatial attention mechanism based on the dependency human contrast sensitivity on spatial frequency. We identify regions in input images, based on underlying spatial frequency where the visual system might be most sensitive to distortions. Based on this prior, we design an attention map that is applied to feature maps in the perceptual loss, helping it to identify regions that are of more perceptual importance. The results will demonstrate that the proposed technique helps improving the correlation of the perceptual loss with human subjective assessment of perceptual quality and also results in a loss which delivers a better perception-distortion trade-off compared to the widely used perceptual loss in CNN based image restoration problems.
http://arxiv.org/abs/1904.00205
Recently, the convolutional neural network has brought impressive improvements for object detection. However, detecting tiny objects in large-scale remote sensing images still remains challenging. First, the extreme large input size makes the existing object detection solutions too slow for practical use. Second, the massive and complex backgrounds cause serious false alarms. Moreover, the ultratiny objects increase the difficulty of accurate detection. To tackle these problems, we propose a unified and self-reinforced network called remote sensing region-based convolutional neural network ($\mathcal{R}^2$-CNN), composing of backbone Tiny-Net, intermediate global attention block, and final classifier and detector. Tiny-Net is a lightweight residual structure, which enables fast and powerful features extraction from inputs. Global attention block is built upon Tiny-Net to inhibit false positives. Classifier is then used to predict the existence of targets in each patch, and detector is followed to locate them accurately if available. The classifier and detector are mutually reinforced with end-to-end training, which further speed up the process and avoid false alarms. Effectiveness of $\mathcal{R}^2$-CNN is validated on hundreds of GF-1 images and GF-2 images that are 18 000 $\times$ 18 192 pixels, 2.0-m resolution, and 27 620 $\times$ 29 200 pixels, 0.8-m resolution, respectively. Specifically, we can process a GF-1 image in 29.4 s on Titian X just with single thread. According to our knowledge, no previous solution can detect the tiny object on such huge remote sensing images gracefully. We believe that it is a significant step toward practical real-time remote sensing systems.
http://arxiv.org/abs/1902.06042
As we interact with the world, for example when we communicate with our colleagues in a large open space or meeting room, we continuously analyse the surrounding environment and, in particular, localise and recognise acoustic events. While we largely take such abilities for granted, they represent a challenging problem for current robots or smart voice assistants as they can be easily fooled by high degree of sound interference in acoustically complex environments. Preventing such failures when using solely audio data is challenging, if not impossible since the algorithms need to take into account wider context and often understand the scene on a semantic level. In this paper, we propose what to our knowledge is the first multi-modal direction of arrival (DoA) of sound, which uses static visual spatial prior providing an auxiliary information about the environment to suppress some of the false DoA detections. We validate our approach on a newly collected real-world dataset, and show that our approach consistently improves over classic DoA baselines
http://arxiv.org/abs/1904.00202