This paper addresses the problem of 3D human pose estimation from single images. While for a long time human skeletons were parameterized and fitted to the observation by satisfying a reprojection error, nowadays researchers directly use neural networks to infer the 3D pose from the observations. However, most of these approaches ignore the fact that a reprojection constraint has to be satisfied and are sensitive to overfitting. We tackle the overfitting problem by ignoring 2D to 3D correspondences. This efficiently avoids a simple memorization of the training data and allows for a weakly supervised training. One part of the proposed reprojection network (RepNet) learns a mapping from a distribution of 2D poses to a distribution of 3D poses using an adversarial training approach. Another part of the network estimates the camera. This allows for the definition of a network layer that performs the reprojection of the estimated 3D pose back to 2D which results in a reprojection loss function. Our experiments show that RepNet generalizes well to unknown data and outperforms state-of-the-art methods when applied to unseen data. Moreover, our implementation runs in real-time on a standard desktop PC.
http://arxiv.org/abs/1902.09868
We propose a novel method for multi-phase segmentation of images based on high-dimensional local feature vectors. While the method was developed for the segmentation of extremely noisy crystal images based on localized Fourier transforms, the resulting framework is not tied to specific feature descriptors. For instance, using local spectral histograms as features, it allows for robust texture segmentation. The segmentation itself is based on the multi-phase Mumford-Shah model. Initializing the high-dimensional mean features directly is computationally too demanding and ill-posed in practice. This is resolved by projecting the features onto a low-dimensional space using principle component analysis. The resulting objective functional is minimized using a convexification and the Chambolle-Pock algorithm. Numerical results are presented, illustrating that the algorithm is very competitive in texture segmentation with state-of-the-art performance on the Prague benchmark and provides new possibilities in crystal segmentation, being robust to extreme noise and requiring no prior knowledge of the crystal structure.
http://arxiv.org/abs/1902.09863
This paper takes a step towards theoretical analysis of the relationship between word embeddings and context embeddings in models such as word2vec. We start from basic probabilistic assumptions on the nature of word vectors, context vectors, and text generation. These assumptions are well supported either empirically or theoretically by the existing literature. Next, we show that under these assumptions the widely-used word-word PMI matrix is approximately a random symmetric Gaussian ensemble. This, in turn, implies that context vectors are reflections of word vectors in approximately half the dimensions. As a direct application of our result, we suggest a theoretically grounded way of tying weights in the SGNS model.
http://arxiv.org/abs/1902.09859
Accurate computer-assisted diagnosis can alleviate the risk of overlooking the diagnosis in a clinical environment. Towards this, as a Data Augmentation (DA) technique, Generative Adversarial Networks (GANs) can synthesize additional training data to handle small/fragmented medical images from various scanners; those images are realistic but completely different from the original ones, filling the data lack in the real image distribution. However, we cannot easily use them to locate the position of disease areas, considering expert physicians’ annotation as time-expensive tasks. Therefore, this paper proposes Conditional Progressive Growing of GANs (CPGGANs), incorporating bounding box conditions into PGGANs to place brain metastases at desired position/size on 256 x 256 Magnetic Resonance (MR) images, for Convolutional Neural Network-based tumor detection; this first GAN-based medical DA using automatic bounding box annotation improves the robustness during training. The results show that CPGGAN-based DA can boost 10% sensitivity in diagnosis with an acceptable amount of additional False Positives—even with physicians’ highly-rough and inconsistent bounding box annotation. Surprisingly, further realistic tumor appearance, achieved with additional normal brain MR images for CPGGAN training, does not contribute to detection performance, while even three expert physicians cannot accurately distinguish them from the real ones in Visual Turing Test.
http://arxiv.org/abs/1902.09856
A 3D point cloud describes the real scene precisely and intuitively.To date how to segment diversified elements in such an informative 3D scene is rarely discussed. In this paper, we first introduce a simple and flexible framework to segment instances and semantics in point clouds simultaneously. Then, we propose two approaches which make the two tasks take advantage of each other, leading to a win-win situation. Specifically, we make instance segmentation benefit from semantic segmentation through learning semantic-aware point-level instance embedding. Meanwhile, semantic features of the points belonging to the same instance are fused together to make more accurate per-point semantic predictions. Our method largely outperforms the state-of-the-art method in 3D instance segmentation along with a significant improvement in 3D semantic segmentation. Code has been made available at: https://github.com/WXinlong/ASIS.
http://arxiv.org/abs/1902.09852
Recently, realistic data augmentation using neural networks especially generative neural networks (GAN) has achieved outstanding results. The communities main research focus is visual image processing. However, automotive cars and robots are equipped with a large suite of sensors to achieve a high redundancy. In addition to others, ultrasonic sensors are often used due to their low-costs and reliable near field distance measuring capabilities. Hence, Pattern recognition needs to be applied to ultrasonic signals as well. Machine Learning requires extensive data sets and those measurements are time-consuming, expensive and not flexible to hardware and environmental changes. On the other hand, there exists no method to simulate those signals deterministically. We present a novel approach for synthetic ultrasonic signal simulation using conditional GANs (cGANs). For the best of our knowledge, we present the first realistic data augmentation for automotive ultrasonics. The performance of cGANs allows us to bring the realistic environment simulation to a new level. By using setup and environmental parameters as condition, the proposed approach is flexible to external influences. Due to the low complexity and time effort for data generation, we outperform other simulation algorithms, such as finite element method. We verify the outstanding accuracy and realism of our method by applying a detailed statistical analysis and comparing the generated data to an extensive amount of measured signals.
http://arxiv.org/abs/1902.09842
Decentralized policies for information gathering are required when multiple autonomous agents are deployed to collect data about a phenomenon of interest without the ability to communicate. Decentralized partially observable Markov decision processes (Dec-POMDPs) are a general, principled model well-suited for such decentralized multiagent decision-making problems. In this paper, we investigate Dec-POMDPs for decentralized information gathering problems. An optimal solution of a Dec-POMDP maximizes the expected sum of rewards over time. To encourage information gathering, we set the reward as a function of the agents’ state information, for example the negative Shannon entropy. We prove that if the reward is convex, then the finite-horizon value function of the corresponding Dec-POMDP is also convex. We propose the first heuristic algorithm for information gathering Dec-POMDPs, and empirically prove its effectiveness by solving problems an order of magnitude larger than previous state-of-the-art.
http://arxiv.org/abs/1902.09840
High performance ultrasonic sensor hardware is mainly used in medical applications. Although, the development in automotive scenarios is towards autonomous driving, the ultrasonic sensor hardware still stays low-cost and low-performance, respectively. To overcome the strict hardware limitations, we propose to use capsule neural networks. By the high classification capability of this network architecture, we can achieve outstanding results for performing a detailed height analysis of detected objects. We apply a novel resorting and reshaping method to feed the neural network with ultrasonic data. This increases classification performance and computation speed. We tested the approach under different environmental conditions to verify that the proposed method is working independent of external parameters that is needed for autonomous driving.
http://arxiv.org/abs/1902.09839
World-class human players have been outperformed in a number of complex two person games (Go, Chess, Checkers) by Deep Reinforcement Learning systems. However, owing to tractability considerations minimax regret of a learning system cannot be evaluated in such games. In this paper we consider simple games (Noughts-and-Crosses and Hexapawn) in which minimax regret can be efficiently evaluated. We use these games to compare Cumulative Minimax Regret for variants of both standard and deep reinforcement learning against two variants of a new Meta-Interpretive Learning system called MIGO. In our experiments all tested variants of both normal and deep reinforcement learning have worse performance (higher cumulative minimax regret) than both variants of MIGO on Noughts-and-Crosses and Hexapawn. Additionally, MIGO’s learned rules are relatively easy to comprehend, and are demonstrated to achieve significant transfer learning in both directions between Noughts-and-Crosses and Hexapawn.
http://arxiv.org/abs/1902.09835
We proposed an end-to-end deep learning-based simultaneous localization and mapping (SLAM) system following conventional visual odometry (VO) pipelines. The proposed method completes the SLAM framework by including tracking, mapping, and sequential optimization networks while training them in an unsupervised manner. Together with the camera pose and depth map, we estimated the observational uncertainty to make our system robust to noises such as dynamic objects. We evaluated our method using public indoor and outdoor datasets. The experiment demonstrated that our method works well in tracking and mapping tasks and performs comparably with other learning-based VO approaches. Notably, the proposed uncertainty modeling and sequential training yielded improved generality in a variety of environments.
http://arxiv.org/abs/1902.09826
Loop-closure detection (LCD) in large non-stationary environments remains an important challenge in robotic visual simultaneous localization and mapping (vSLAM). To reduce computational and perceptual complexity, it is helpful if a vSLAM system has the ability to perform image change detection (ICD). Unlike previous applications of ICD, time-critical vSLAM applications cannot assume an offline background modeling stage, or rely on maintenance-intensive background models. To address this issue, we introduce a novel maintenance-free ICD framework that requires no background modeling. Specifically, we demonstrate that LCD can be reused as the main process for ICD with minimal extra cost. Based on these concepts, we develop a novel vSLAM component that enables simultaneous LCD and ICD. ICD experiments based on challenging cross-season LCD scenarios validate the efficacy of the proposed method.
http://arxiv.org/abs/1902.09822
Objective image quality assessment (IQA) is imperative in the current multimedia-intensive world, in order to assess the visual quality of an image at close to a human level of ability. Many~parameters such as color intensity, structure, sharpness, contrast, presence of an object, etc., draw human attention to an image. Psychological vision research suggests that human vision is biased to the center area of an image and display screen. As a result, if the center part contains any visually salient information, it draws human attention even more and any distortion in that part will be better perceived than other parts. To the best of our knowledge, previous IQA methods have not considered this fact. In this paper, we propose a full reference image quality assessment (FR-IQA) approach using visual saliency and contrast; however, we give extra attention to the center by increasing the sensitivity of the similarity maps in that region. We evaluated our method on three large-scale popular benchmark databases used by most of the current IQA researchers (TID2008, CSIQ~and LIVE), having a total of 3345 distorted images with 28~different kinds of distortions. Our~method is compared with 13 state-of-the-art approaches. This comparison reveals the stronger correlation of our method with human-evaluated values. The prediction-of-quality score is consistent for distortion specific as well as distortion independent cases. Moreover, faster processing makes it applicable to any real-time application. The MATLAB code is publicly available to test the algorithm and can be found online at this http URL
http://arxiv.org/abs/1812.11163
The key challenge of generative Visual Dialogue (VD) systems is to respond to human queries with informative answers in natural and contiguous conversation flow. Traditional Maximum Likelihood Estimation (MLE)-based methods only learn from positive responses but ignore the negative responses, and consequently tend to yield safe or generic responses. To address this issue, we propose a novel training scheme in conjunction with weighted likelihood estimation (WLE) method. Furthermore, an adaptive multi-modal reasoning module is designed, to accommodate various dialogue scenarios automatically and select relevant information accordingly. The experimental results on the VisDial benchmark demonstrate the superiority of our proposed algorithm over other state-of-the-art approaches, with an improvement of 5.81% on recall@10.
http://arxiv.org/abs/1902.09818
Graph Convolutional Networks (GCNs) have proved to be a most powerful architecture in aggregating local neighborhood information for individual graph nodes. Low-rank proximities and node features are successfully leveraged in existing GCNs, however, attributes that graph links may carry are commonly ignored, as almost all of these models simplify graph links into binary or scalar values describing node connectedness. In our paper instead, links are reverted to hypostatic relationships between entities with descriptional attributes. We propose GCN-LASE (GCN with Link Attributes and Sampling Estimation), a novel GCN model taking both node and link attributes as inputs. To adequately captures the interactions between link and node attributes, their tensor product is used as neighbor features, based on which we define several graph kernels and further develop according architectures for LASE. Besides, to accelerate the training process, the sum of features in entire neighborhoods are estimated through Monte Carlo method, with novel sampling strategies designed for LASE to minimize the estimation variance. Our experiments show that LASE outperforms strong baselines over various graph datasets, and further experiments corroborate the informativeness of link attributes and our model’s ability of adequately leveraging them.
http://arxiv.org/abs/1902.09817
Example synthesis is one of the leading methods to tackle the problem of few-shot learning, where only a small number of samples per class are available. However, current synthesis approaches only address the scenario of a single category label per image. In this work, we propose a novel technique for synthesizing samples with multiple labels for the (yet unhandled) multi-label few-shot classification scenario. We propose to combine pairs of given examples in feature space, so that the resulting synthesized feature vectors will correspond to examples whose label sets are obtained through certain set operations on the label sets of the corresponding input pairs. Thus, our method is capable of producing a sample containing the intersection, union or set-difference of labels present in two input samples. As we show, these set operations generalize to labels unseen during training. This enables performing augmentation on examples of novel categories, thus, facilitating multi-label few-shot classifier learning. We conduct numerous experiments showing promising results for the label-set manipulation capabilities of the proposed approach, both directly (using the classification and retrieval metrics), and in the context of performing data augmentation for multi-label few-shot learning. We propose a benchmark for this new and challenging task and show that our method compares favorably to all the common baselines.
http://arxiv.org/abs/1902.09811
Recurrent convolution (RC) shares the same convolutional kernels and unrolls them multiple steps, which is originally proposed to model time-space signals. We argue that RC can be viewed as a model compression strategy for deep convolutional neural networks. RC reduces the redundancy across layers. However, the performance of an RC network is not satisfactory if we directly unroll the same kernels multiple steps. We propose a simple yet effective variant which improves the RC networks: the batch normalization layers of an RC module are learned independently (not shared) for different unrolling steps. Moreover, we verify that RC can perform cost-adjustable inference which is achieved by varying its unrolling steps. We learn double independent BN layers for cost-adjustable RC networks, i.e. independent w.r.t both the unrolling steps of current cell and upstream cell. We provide insights on why the proposed method works successfully. Experiments on both image classification and image denoise demonstrate the effectiveness of our method.
http://arxiv.org/abs/1902.09809
Capturing the meaning of sentences has long been a challenging task. Current models tend to apply linear combinations of word features to conduct semantic composition for bigger-granularity units e.g. phrases, sentences, and documents. However, the semantic linearity does not always hold in human language. For instance, the meaning of the phrase ivory tower' can not be deduced by linearly combining the meanings of ivory’ and `tower’. To address this issue, we propose a new framework that models different levels of semantic units (e.g. sememe, word, sentence, and semantic abstraction) on a single \textit{Semantic Hilbert Space}, which naturally admits a non-linear semantic composition by means of a complex-valued vector word representation. An end-to-end neural network~\footnote{https://github.com/wabyking/qnn} is proposed to implement the framework in the text classification task, and evaluation results on six benchmarking text classification datasets demonstrate the effectiveness, robustness and self-explanation power of the proposed model. Furthermore, intuitive case studies are conducted to help end users to understand how the framework works.
http://arxiv.org/abs/1902.09802
In human-computer interaction, it is important to accurately estimate the hand pose especially fingertips. However, traditional approaches for fingertip localization mainly rely on depth images and thus suffer considerably from the noise and missing values. Instead of depth images, stereo images can also provide 3D information of hands and promote 3D hand pose estimation. There are nevertheless limitations on the dataset size, global viewpoints, hand articulations and hand shapes in the publicly available stereo-based hand pose datasets. To mitigate these limitations and promote further research on hand pose estimation from stereo images, we propose a new large-scale binocular hand pose dataset called THU-Bi-Hand, offering a new perspective for fingertip localization. In the THU-Bi-Hand dataset, there are 447k pairs of stereo images of different hand shapes from 10 subjects with accurate 3D location annotations of the wrist and five fingertips. Captured with minimal restriction on the range of hand motion, the dataset covers large global viewpoint space and hand articulation space. To better present the performance of fingertip localization on THU-Bi-Hand, we propose a novel scheme termed Bi-stream Pose Guided Region Ensemble Network (Bi-Pose-REN). It extracts more representative feature regions around joint points in the feature maps under the guidance of the previously estimated pose. The feature regions are integrated hierarchically according to the topology of hand joints to regress the refined hand pose. Bi-Pose-REN and several existing methods are evaluated on THU-Bi-Hand so that benchmarks are provided for further research. Experimental results show that our new method has achieved the best performance on THU-Bi-Hand.
http://arxiv.org/abs/1902.09795
There are many facts affecting human face recognition, such as pose, occlusion, illumination, age, etc. First and foremost are large pose and occlusion problems, which can even result in more than 10% performance degradation. Pose-invariant feature representation and face frontalization with generative adversarial networks (GAN) have been widely used to solve the pose problem. However, the synthesis and recognition of occlusive but profile faces is still an uninvestigated problem. To address this issue, in this paper, we aim to contribute an effective solution on how to recognize occlusive but profile faces, even with facial keypoint region (e.g. eyes, nose, etc.) corrupted. Specifically, we propose a boosting Generative Adversarial Network (BoostGAN) for de-occlusion, frontalization, and recognition of faces. Upon the assumption that facial occlusion is partial and incomplete, multiple patch occluded images are fed as inputs for knowledge boosting, such as identity and texture information. A new aggregation structure composed of a deep GAN for coarse face synthesis and a shallow boosting net for fine face generation is further designed. Exhaustive experiments demonstrate that the proposed approach not only presents clear perceptual photo-realistic results but also shows state-of-the-art recognition performance for occlusive but profile faces.
http://arxiv.org/abs/1902.09782
The need for tree structure modelling on top of sequence modelling is an open issue in neural dependency parsing. We investigate the impact of adding a tree layer on top of a sequential model by recursively composing subtree representations (composition) in a transition-based parser that uses features extracted by a BiLSTM. Composition seems superfluous with such a model, suggesting that BiLSTMs capture information about subtrees. We perform model ablations to tease out the conditions under which composition helps. When ablating the backward LSTM, performance drops and composition does not recover much of the gap. When ablating the forward LSTM, performance drops less dramatically and composition recovers a substantial part of the gap, indicating that a forward LSTM and composition capture similar information. We take the backward LSTM to be related to lookahead features and the forward LSTM to the rich history-based features both crucial for transition-based parsers. To capture history-based information, composition is better than a forward LSTM on its own, but it is even better to have a forward LSTM as part of a BiLSTM. We correlate results with language properties, showing that the improved lookahead of a backward LSTM is especially important for head-final languages.
http://arxiv.org/abs/1902.09781
This study provides a systematic review of the recent advances in designing the intelligent tutoring robot (ITR), and summarises the status quo of applying artificial intelligence (AI) techniques. We first analyse the environment of the ITR and propose a relationship model for describing interactions of ITR with the students, the social milieu and the curriculum. Then, we transform the relationship model into the perception-planning-action model for exploring what AI techniques are suitable to be applied in the ITR. This article provides insights on promoting human-robot teaching-learning process and AI-assisted educational techniques, illustrating the design guidelines and future research perspectives in intelligent tutoring robots.
http://arxiv.org/abs/1903.03414
Single-image piece-wise planar 3D reconstruction aims to simultaneously segment plane instances and recover 3D plane parameters from an image. Most recent approaches leverage convolutional neural networks (CNNs) and achieve promising results. However, these methods are limited to detecting a fixed number of planes with certain learned order. To tackle this problem, we propose a novel two-stage method based on associative embedding, inspired by its recent success in instance segmentation. In the first stage, we train a CNN to map each pixel to an embedding space where pixels from the same plane instance have similar embeddings. Then, the plane instances are obtained by grouping the embedding vectors in planar regions via an efficient mean shift clustering algorithm. In the second stage, we estimate the parameter for each plane instance by considering both pixel-level and instance-level consistencies. With the proposed method, we are able to detect an arbitrary number of planes. Extensive experiments on public datasets validate the effectiveness and efficiency of our method. Furthermore, our method runs at 30 fps at the testing time, thus could facilitate many real-time applications such as visual SLAM and human-robot interaction. Code is available at https://github.com/svip-lab/PlanarReconstruction.
http://arxiv.org/abs/1902.09777
The image, question (combined with the history for de-referencing), and the corresponding answer are three vital components of visual dialog. Classical visual dialog systems integrate the image, question, and history to search for or generate the best matched answer, and so, this approach significantly ignores the role of the answer. In this paper, we devise a novel image-question-answer synergistic network to value the role of the answer for precise visual dialog. We extend the traditional one-stage solution to a two-stage solution. In the first stage, candidate answers are coarsely scored according to their relevance to the image and question pair. Afterward, in the second stage, answers with high probability of being correct are re-ranked by synergizing with image and question. On the Visual Dialog v1.0 dataset, the proposed synergistic network boosts the discriminative visual dialog model to achieve a new state-of-the-art of 57.88\% normalized discounted cumulative gain. A generative visual dialog model equipped with the proposed technique also shows promising improvements.
http://arxiv.org/abs/1902.09774
This paper proposes a data-efficient, semi-supervised, two-pass framework for segmenting bird vocalizations. The framework utilizes a binary classification model to categorize frames of an input audio recording into the background or bird vocalization. The first pass of the framework automatically generates training labels from the input recording itself, while model training and classification is done during the second pass. The proposed framework utilizes a reference directional model for obtaining a feature representation called directional embeddings (DE). This reference directional model acts as an acoustic model for bird vocalizations and is obtained using the mixtures of Von-Mises Fisher distribution (moVMF). The proposed DE space only contains information about bird vocalizations, while no information about the background disturbances is reflected. The framework employs supervised information only for obtaining the reference directional model and avoids the background modeling. Hence, it can be regarded as semi-supervised in nature. The proposed framework is tested on approximately 79000 vocalizations of seven different bird species. The performance of the framework is also analyzed in the presence of noise at different SNRs. Experimental results convey that the proposed framework performs better than the existing bird vocalization segmentation methods.
http://arxiv.org/abs/1902.09765
We propose a 3D object detection method for autonomous driving by fully exploiting the sparse and dense, semantic and geometry information in stereo imagery. Our method, called Stereo R-CNN, extends Faster R-CNN for stereo inputs to simultaneously detect and associate object in left and right images. We add extra branches after stereo Region Proposal Network (RPN) to predict sparse keypoints, viewpoints, and object dimensions, which are combined with 2D left-right boxes to calculate a coarse 3D object bounding box. We then recover the accurate 3D bounding box by a region-based photometric alignment using left and right RoIs. Our method does not require depth input and 3D position supervision, however, outperforms all existing fully supervised image-based methods. Experiments on the challenging KITTI dataset show that our method outperforms the state-of-the-art stereo-based method by around 30% AP on both 3D detection and 3D localization tasks. Code will be made publicly available.
http://arxiv.org/abs/1902.09738
The recent direction of unpaired image-to-image translation is on one hand very exciting as it alleviates the big burden in obtaining label-intensive pixel-to-pixel supervision, but it is on the other hand not fully satisfactory due to the presence of artifacts and degenerated transformations. In this paper, we take a manifold view of the problem by introducing a smoothness term over the sample graph to attain harmonic functions to enforce consistent mappings during the translation. We develop HarmonicGAN to learn bi-directional translations between the source and the target domains. With the help of similarity-consistency, the inherent self-consistency property of samples can be maintained. Distance metrics defined on two types of features including histogram and CNN are exploited. Under an identical problem setting as CycleGAN, without additional manual inputs and only at a small training-time cost, HarmonicGAN demonstrates a significant qualitative and quantitative improvement over the state of the art, as well as improved interpretability. We show experimental results in a number of applications including medical imaging, object transfiguration, and semantic labeling. We outperform the competing methods in all tasks, and for a medical imaging task in particular our method turns CycleGAN from a failure to a success, halving the mean-squared error, and generating images that radiologists prefer over competing methods in 95% of cases.
http://arxiv.org/abs/1902.09727
Reward functions are often misspecified. An agent optimizing an incorrect reward function can change its environment in large, undesirable, and potentially irreversible ways. Work on impact measurement seeks a means of identifying (and thereby avoiding) large changes to the environment. We propose a novel impact measure which induces conservative, effective behavior across a range of situations. The approach attempts to preserve the attainable utility of auxiliary objectives. We evaluate our proposal on an array of benchmark tasks and show that it matches or outperforms relative reachability, the state-of-the-art in impact measurement.
http://arxiv.org/abs/1902.09725
Writing style is a combination of consistent decisions at different levels of language production including lexical, syntactic, and structural associated to a specific author (or author groups). While lexical-based models have been widely explored in style-based text classification, relying on context makes the model less scalable when dealing with heterogeneous data comprised of various topics. On the other hand, syntactic models which are context-independent, are more robust against topic variance. In this paper, we introduce a syntactic recurrent neural network to encode the syntactic patterns of a document in a hierarchical structure. The model first learns the syntactic representation of sentences from the sequence of part-of-speech tags. For this purpose, we exploit both convolutional filters and long short-term memories to investigate the short-term and long-term dependencies of part-of-speech tags in the sentences. Subsequently, the syntactic representations of sentences are aggregated into document representation using recurrent neural networks. Our experimental results on PAN 2012 dataset for authorship attribution task shows that syntactic recurrent neural network outperforms the lexical model with the identical architecture by approximately 14% in terms of accuracy.
http://arxiv.org/abs/1902.09723
Deep ConvNets have shown great performance for single-label image classification (e.g. ImageNet), but it is necessary to move beyond the single-label classification task because pictures of everyday life are inherently multi-label. Multi-label classification is a more difficult task than single-label classification because both the input images and output label spaces are more complex. Furthermore, collecting clean multi-label annotations is more difficult to scale-up than single-label annotations. To reduce the annotation cost, we propose to train a model with partial labels i.e. only some labels are known per image. We first empirically compare different labeling strategies to show the potential for using partial labels on multi-label datasets. Then to learn with partial labels, we introduce a new classification loss that exploits the proportion of known labels per example. Our approach allows the use of the same training settings as when learning with all the annotations. We further explore several curriculum learning based strategies to predict missing labels. Experiments are performed on three large-scale multi-label datasets: MS COCO, NUS-WIDE and Open Images.
http://arxiv.org/abs/1902.09720
We propose a method to create document representations that reflect their internal structure. We modify Tree-LSTMs to hierarchically merge basic elements like words and sentences into blocks of increasing complexity. Our Structure Tree-LSTM implements a hierarchical attention mechanism over individual components and combinations thereof. We thus emphasize the usefulness of Tree-LSTMs for texts larger than a sentence. We show that structure-aware encoders can be used to improve the performance of document classification. We demonstrate that our method is resilient to changes to the basic building blocks, as it performs well with both sentence and word embeddings. The Structure Tree-LSTM outperforms all the baselines on two datasets when structural clues like sections are available, but also in the presence of mere paragraphs. On a third dataset from the medical domain, our model achieves competitive performance with the state of the art. This result shows the Structure Tree-LSTM can leverage dependency relations other than text structure, such as a set of reports on the same patient.
http://arxiv.org/abs/1902.09713
The past few years have witnessed great success in applying deep learning to enhance the quality of compressed image/video. The existing approaches mainly focus on enhancing the quality of a single frame, not considering the similarity between consecutive frames. Since heavy fluctuation exists across compressed video frames as investigated in this paper, frame similarity can be utilized for quality enhancement of low-quality frames by using their neighboring high-quality frames. This task can be seen as Multi-Frame Quality Enhancement (MFQE). Accordingly, this paper proposes an MFQE approach for compressed video, as the first attempt in this direction. In our approach, we firstly develop a Bidirectional Long Short-Term Memory (BiLSTM) based detector to locate Peak Quality Frames (PQFs) in compressed video. Then, a novel Multi-Frame Convolutional Neural Network (MF-CNN) is designed to enhance the quality of compressed video, in which the non-PQF and its nearest two PQFs are the input. In MF-CNN, motion between the non-PQF and PQFs is compensated by a motion compensation subnet. Subsequently, a quality enhancement subnet fuses the non-PQF and compensated PQFs, and then reduces the compression artifacts of the non-PQF. Finally, experiments validate the effectiveness and generalization ability of our MFQE approach in advancing the state-of-the-art quality enhancement of compressed video. The code of our MFQE approach is available at https://github.com/RyanXingQL/MFQE2.0.git.
http://arxiv.org/abs/1902.09707
Variational autoencoders~(VAEs) have shown a promise in data-driven conversation modeling. However, most VAE conversation models match the approximate posterior distribution over the latent variables to a simple prior such as standard normal distribution, thereby restricting the generated responses to a relatively simple (e.g., unimodal) scope. In this paper, we propose DialogWAE, a conditional Wasserstein autoencoder~(WAE) specially designed for dialogue modeling. Unlike VAEs that impose a simple distribution over the latent variables, DialogWAE models the distribution of data by training a GAN within the latent variable space. Specifically, our model samples from the prior and posterior distributions over the latent variables by transforming context-dependent random noise using neural networks and minimizes the Wasserstein distance between the two distributions. We further develop a Gaussian mixture prior network to enrich the latent space. Experiments on two popular datasets show that DialogWAE outperforms the state-of-the-art approaches in generating more coherent, informative and diverse responses.
http://arxiv.org/abs/1805.12352
It is crucial to generate crafted SAT formulas with predefined solutions for the testing and development of SAT solvers since many SAT formulas from real-world applications have solutions. Although some generating algorithms have been proposed to generate SAT formulas with predefined solutions, community structures of SAT formulas are not considered. We propose a 3-SAT formula generating algorithm that not only guarantees the existence of a predefined solution, but also simultaneously considers community structures and clause distributions. The proposed 3-SAT formula generating algorithm controls the quality of community structures through controlling (1) the number of clauses whose variables have a common community, which we call intra-community clauses, and (2) the number of variables that only belong to one community, which we call intra-community variables. To study the combined effect of community structures and clause distributions on the hardness of SAT formulas, we measure solving runtimes of two solvers, gluHack (a leading CDCL solver) and CPSparrow (a leading SLS solver), on the generated SAT formulas under different groups of parameter settings. Through extensive experiments, we obtain some noteworthy observations on the SAT formulas generated by the proposed algorithm: (1) The community structure has little or no effects on the hardness of SAT formulas with regard to CPSparrow but a strong effect with regard to gluHack. (2) Only when the proportion of true literals in a SAT formula in terms of the predefined solution is 0.5, SAT formulas are hard-to-solve with regard to gluHack; when this proportion is below 0.5, SAT formulas are hard-to-solve with regard to CPSparrow. (3) When the ratio of the number of clauses to that of variables is around 4.25, the SAT formulas are hard-to-solve with regard to both gluHack and CPSparrow.
http://arxiv.org/abs/1902.09706
We propose a developmental approach that allows a robot to interpret and describe the actions of human agents by reusing previous experience. The robot first learns the association between words and object affordances by manipulating the objects in its environment. It then uses this information to learn a mapping between its own actions and those performed by a human in a shared environment. It finally fuses the information from these two models to interpret and describe human actions in light of its own experience. In our experiments, we show that the model can be used flexibly to do inference on different aspects of the scene. We can predict the effects of an action on the basis of object properties. We can revise the belief that a certain action occurred, given the observed effects of the human action. In an early action recognition fashion, we can anticipate the effects when the action has only been partially observed. By estimating the probability of words given the evidence and feeding them into a pre-defined grammar, we can generate relevant descriptions of the scene. We believe that this is a step towards providing robots with the fundamental skills to engage in social collaboration with humans.
http://arxiv.org/abs/1902.09705
A Generative Adversarial Network (GAN) with generator $G$ trained to model the prior of images has been shown to perform better than sparsity-based regularizers in ill-posed inverse problems. In this work, we propose a new method of deploying a GAN-based prior to solve linear inverse problems using projected gradient descent (PGD). Our method learns a network-based projector for use in the PGD algorithm, eliminating the need for expensive computation of the Jacobian of $G$. Experiments show that our approach provides a speed-up of $30\text{-}40\times$ over earlier GAN-based recovery methods for similar accuracy in compressed sensing. Our main theoretical result is that if the measurement matrix is moderately conditioned for range($G$) and the projector is $\delta$-approximate, then the algorithm is guaranteed to reach $O(\delta)$ reconstruction error in $O(log(1/\delta))$ steps in the low noise regime. Additionally, we propose a fast method to design such measurement matrices for a given $G$. Extensive experiments demonstrate the efficacy of this method by requiring $5\text{-}10\times$ fewer measurements than random Gaussian measurement matrices for comparable recovery performance.
https://arxiv.org/abs/1902.09698
We introduce a method to produce multilingual contextual word representations by training a single language model on text from multiple languages. Our method combines the advantages of contextual word representations with those of multilingual representation learning. We produce language models from dissimilar language pairs (English/Arabic and English/Chinese) and use them in dependency parsing, semantic role labeling, and named entity recognition, with comparisons to monolingual and non-contextual variants. Our results provide further support for polyglot learning, in which representations are shared across multiple languages.
http://arxiv.org/abs/1902.09697
Effective network slicing requires an infrastructure/network provider to deal with the uncertain demand and real-time dynamics of network resource requests. Another challenge is the combinatorial optimization of numerous resources, e.g., radio, computing, and storage. This article develops an optimal and fast real-time resource slicing framework that maximizes the long-term return of the network provider while taking into account the uncertainty of resource demand from tenants. Specifically, we first propose a novel system model which enables the network provider to effectively slice various types of resources to different classes of users under separate virtual slices. We then capture the real-time arrival of slice requests by a semi-Markov decision process. To obtain the optimal resource allocation policy under the dynamics of slicing requests, e.g., uncertain service time and resource demands, a Q-learning algorithm is often adopted in the literature. However, such an algorithm is notorious for its slow convergence, especially for problems with large state/action spaces. This makes Q-learning practically inapplicable to our case in which multiple resources are simultaneously optimized. To tackle it, we propose a novel network slicing approach with an advanced deep learning architecture, called deep dueling that attains the optimal average reward much faster than the conventional Q-learning algorithm. This property is especially desirable to cope with real-time resource requests and the dynamic demands of users. Extensive simulations show that the proposed framework yields up to 40% higher long-term average return while being few thousand times faster, compared with state of the art network slicing approaches.
http://arxiv.org/abs/1902.09696
In the area of data classification, the different classifiers have been developed by its own strengths and weaknesses. Among these classifiers, we propose a method which is based on the maximum margin between two classes. One of the main challenges in this area is dealt with noisy data. In this paper, our aim is to optimize the method of large margin classifier based on hyperdisk (LMC-HD) and incorporate it into quasi-support vector data description (QSVDD) method. In the proposed method, the bounding hypersphere is calculated based on the QSVDD method. So our convex class model is more robust in compared with support vector machine (SVM) and less tight than LMC-HD. Applying this idea causes the reduction of the impact of the noisy data set in classification. Large margin classifiers aim to maximize the margin and minimizing the risk. Sine our proposed method ignores the effect of outliers and noises, so this method has the widest margin compared with other large margin classifiers. In the end, we compare our proposed method with other popular large margin classifiers by the experiments on a set of standard data which indicates our results are more efficient than the others.
http://arxiv.org/abs/1902.09692
In recent years, with the development of the marine industry, navigation environment becomes more complicated. Some artificial intelligence technologies, such as computer vision, can recognize, track and count the sailing ships to ensure the maritime security and facilitates the management for Smart Ocean System. Aiming at the scaling problem and boundary effect problem of traditional correlation filtering methods, we propose a self-selective correlation filtering method based on box regression (BRCF). The proposed method mainly include: 1) A self-selective model with negative samples mining method which effectively reduces the boundary effect in strengthening the classification ability of classifier at the same time; 2) A bounding box regression method combined with a key points matching method for the scale prediction, leading to a fast and efficient calculation. The experimental results show that the proposed method can effectively deal with the problem of ship size changes and background interference. The success rates and precisions were higher than Discriminative Scale Space Tracking (DSST) by over 8 percentage points on the marine traffic dataset of our laboratory. In terms of processing speed, the proposed method is higher than DSST by nearly 22 Frames Per Second (FPS).
http://arxiv.org/abs/1902.09690
Video frame synthesis is an active computer vision problem which has applications in video compression, streaming, editing, and understanding. In this work, we present a computational high speed video synthesis framework. Our framework takes as inputs two types of data streams: an intensity frame stream and a neuromorphic event stream, which consists of asynchronous bipolar “events” which encode brightness variations over time at over 1000 fps. We introduce an algorithm to recover a space-time video from these two modes of observations. We factor the reconstruction into a physical model-based reconstruction (PBR) process and a residual denoising process. We use a differentiable model to approximate the physical sensing process, which enables stochastic gradient descent optimization using automatic differentiation. Residual errors in PBR reconstruction are further reduced by training a residual denoiser to remove reconstruction artifacts. The video output from our reconstruction algorithm has both high frame rate and well-recovered spatial features. Our framework is capable of handling challenging scenes that include fast motion and strong occlusions.
http://arxiv.org/abs/1902.09680
Background and Significance: Selecting cohorts for a clinical trial typically requires costly and time-consuming manual chart reviews resulting in poor participation. To help automate the process, National NLP Clinical Challenges (N2C2) conducted a shared challenge by defining 13 criteria for clinical trial cohort selection and by providing training and test datasets. This research was motivated by the N2C2 challenge. Methods: We broke down the task into 13 independent subtasks corresponding to each criterion and implemented subtasks using rules or a supervised machine learning model. Each task critically depended on knowledge resources in the form of task-specific lexicons, for which we developed a novel model-driven approach. The approach allowed us to first expand the lexicon from a seed set and then remove noise from the list, thus improving the accuracy. Results: Our system achieved an overall F measure of 0.9003 at the challenge, and was statistically tied for the first place out of 45 participants. The model-driven lexicon development and further debugging the rules/code on the training set improved overall F measure to 0.9140, overtaking the best numerical result at the challenge. Discussion: Cohort selection, like phenotype extraction and classification, is amenable to rule-based or simple machine learning methods, however, the lexicons involved, such as medication names or medical terms referring to a medical problem, critically determine the overall accuracy. Automated lexicon development has the potential for scalability and accuracy.
http://arxiv.org/abs/1902.09674
As offensive content has become pervasive in social media, there has been much research on identifying potentially offensive messages. Previous work in this area, however, did not consider the problem as a whole, but rather focused on detecting very specific types of offensive content, e.g., hate speech, cyberbulling, or cyber-aggression. In contrast, here we target several different kinds of offensive content. In particular, we propose to model the task hierarchically, identifying the type and the target of offensive messages in social media. We use the Offensive Language Identification Dataset (OLID), a new dataset with a fine-grained three-layer annotation scheme compiled specifically for this purpose. OLID, which we make publicly available, contains tweets annotated for offensive content. We discuss the main similarities and differences of this dataset compared to other datasets for hate speech identification, aggression detection, and similar tasks. We also evaluate the data with a number of classification methods for this task.
http://arxiv.org/abs/1902.09666
Robotic platforms are emerging as a timely and cost-efficient tool for exploration and monitoring. However, an open challenge is planning missions for robust, efficient data acquisition in complex environments. To address this issue, we introduce an informative planning framework for active sensing scenarios that accounts for the robot pose uncertainty. Our strategy exploits a Gaussian Process model to capture a target environmental field given the uncertainty on its inputs. This allows us to maintain robust maps, which are used for planning information-rich trajectories in continuous space. A key aspect of our method is a new utility function that couples the localization and field mapping objectives, enabling us to trade-off exploration against exploitation in a principled way. Extensive simulations show that our approach outperforms existing strategies, with reductions of up to 45.1% and 6.3% in mean pose uncertainty and map error. We demonstrate a proof of concept in an indoor temperature mapping scenario.
http://arxiv.org/abs/1902.09660
Batch Normalization (BN) is a common technique used to speed-up and stabilize training. On the other hand, the learnable parameters of BN are commonly used in conditional Generative Adversarial Networks (cGANs) for representing class-specific information using conditional Batch Normalization (cBN). In this paper we propose to generalize both BN and cBN using a Whitening and Coloring based batch normalization. We show that our conditional Coloring can represent categorical conditioning information which largely helps the cGAN qualitative results. Moreover, we show that full-feature whitening is important in a general GAN scenario in which the training process is known to be highly unstable. We test our approach on different datasets and using different GAN networks and training protocols, showing a consistent improvement in all the tested frameworks. Our CIFAR-10 conditioned results are higher than all previous works on this dataset.
https://arxiv.org/abs/1806.00420
Lesions characterized by computed tomography (CT) scans, are arguably often elliptical objects. However, current lesion detection systems are predominantly adopted from the popular Region Proposal Networks (RPNs) that only propose bounding boxes without fully leveraging the elliptical geometry of lesions. In this paper, we present Gaussian Proposal Networks (GPNs), a novel extension to RPNs, to detect lesion bounding ellipses. Instead of directly regressing the rotation angle of the ellipse as the common practice, GPN represents bounding ellipses as 2D Gaussian distributions on the image plain and minimizes the Kullback-Leibler (KL) divergence between the proposed Gaussian and the ground truth Gaussian for object localization. We show the KL divergence loss approximately incarnates the regression loss in the RPN framework when the rotation angle is 0. Experiments on the DeepLesion dataset show that GPN significantly outperforms RPN for lesion bounding ellipse detection thanks to lower localization error. GPN is open sourced at https://github.com/baidu-research/GPN
http://arxiv.org/abs/1902.09658
We present a method that learns to integrate temporal information, from a learned dynamics model, with ambiguous visual information, from a learned vision model, in the context of interacting agents. Our method is based on a graph-structured variational recurrent neural network (Graph-VRNN), which is trained end-to-end to infer the current state of the (partially observed) world, as well as to forecast future states. We show that our method outperforms various baselines on two sports datasets, one based on real basketball trajectories, and one generated by a soccer game engine.
http://arxiv.org/abs/1902.09641
Recent advances in neural architecture search (NAS) demand tremendous computational resources. This makes it difficult to reproduce experiments and imposes a barrier-to-entry to researchers without access to large-scale computation. We aim to ameliorate these problems by introducing NAS-Bench-101, the first public architecture dataset for NAS research. To build NAS-Bench-101, we carefully constructed a compact, yet expressive, search space, exploiting graph isomorphisms to identify 423k unique convolutional architectures. We trained and evaluated all of these architectures multiple times on CIFAR-10 and compiled the results into a large dataset. All together, NAS-Bench-101 contains the metrics of over 5 million models, the largest dataset of its kind thus far. This allows researchers to evaluate the quality of a diverse range of models in milliseconds by querying the pre-computed dataset. We demonstrate its utility by analyzing the dataset as a whole and by benchmarking a range of architecture optimization algorithms.
https://arxiv.org/abs/1902.09635
Interest in image-to-image translation has grown substantially in recent years with the success of unsupervised models based on the cycle-consistency assumption. The achievements of these models have been limited to a particular subset of domains where this assumption yields good results, namely homogeneous domains that are characterized by style or texture differences. We tackle the challenging problem of image-to-image translation where the domains are defined by high-level shapes and contexts, as well as including significant clutter and heterogeneity. For this purpose, we introduce a novel GAN based on preserving intra-domain vector transformations in a latent space learned by a siamese network. The traditional GAN system introduced a discriminator network to guide the generator into generating images in the target domain. To this two-network system we add a third: a siamese network that guides the generator so that each original image shares semantics with its generated version. With this new three-network system, we no longer need to constrain the generators with the ubiquitous cycle-consistency restraint. As a result, the generators can learn mappings between more complex domains that differ from each other by large differences - not just style or texture.
http://arxiv.org/abs/1902.09631
Intersection over Union (IoU) is the most popular evaluation metric used in the object detection benchmarks. However, there is a gap between optimizing the commonly used distance losses for regressing the parameters of a bounding box and maximizing this metric value. The optimal objective for a metric is the metric itself. In the case of axis-aligned 2D bounding boxes, it can be shown that $IoU$ can be directly used as a regression loss. However, $IoU$ has a plateau making it infeasible to optimize in the case of non-overlapping bounding boxes. In this paper, we address the weaknesses of $IoU$ by introducing a generalized version as both a new loss and a new metric. By incorporating this generalized $IoU$ ($GIoU$) as a loss into the state-of-the art object detection frameworks, we show a consistent improvement on their performance using both the standard, $IoU$ based, and new, $GIoU$ based, performance measures on popular object detection benchmarks such as PASCAL VOC and MS COCO.
http://arxiv.org/abs/1902.09630
Insects and hummingbirds exhibit extraordinary flight capabilities and can simultaneously master seemingly conflicting goals: stable hovering and aggressive maneuvering, unmatched by small scale man-made vehicles. Flapping Wing Micro Air Vehicles (FWMAVs) hold great promise for closing this performance gap. However, design and control of such systems remain challenging due to various constraints. Here, we present an open source high fidelity dynamic simulation for FWMAVs to serve as a testbed for the design, optimization and flight control of FWMAVs. For simulation validation, we recreated the hummingbird-scale robot developed in our lab in the simulation. System identification was performed to obtain the model parameters. The force generation, open-loop and closed-loop dynamic response between simulated and experimental flights were compared and validated. The unsteady aerodynamics and the highly nonlinear flight dynamics present challenging control problems for conventional and learning control algorithms such as Reinforcement Learning. The interface of the simulation is fully compatible with OpenAI Gym environment. As a benchmark study, we present a linear controller for hovering stabilization and a Deep Reinforcement Learning control policy for goal-directed maneuvering. Finally, we demonstrate direct simulation-to-real transfer of both control policies onto the physical robot, further demonstrating the fidelity of the simulation.
http://arxiv.org/abs/1902.09628