Exploring the potential of GANs for unsupervised disentanglement learning, this paper proposes a novel framework called OOGAN. While previous work mostly attempts to tackle disentanglement learning through VAE and seeks to minimize the Total Correlation (TC) objective with various sorts of approximation methods, we show that GANs have a natural advantage in disentangling with a straightforward latent variable sampling method. Furthermore, we provide a brand-new perspective on designing the structure of the generator and discriminator, demonstrating that a minor structural change and an orthogonal regularization on model weights entails improved disentanglement learning. Our experiments on several visual datasets confirm the effectiveness and superiority of this approach.
http://arxiv.org/abs/1905.10836
Automatic identification of brain lesions from magnetic resonance imaging (MRI) scans of stroke survivors would be a useful aid in patient diagnosis and treatment planning. We propose a multi-modal multi-path convolutional neural network system for automating stroke lesion segmentation. Our system has nine end-to-end UNets that take as input 2-dimensional (2D) slices and examines all three planes with three different normalizations. Outputs from these nine total paths are concatenated into a 3D volume that is then passed to a 3D convolutional neural network to output a final lesion mask. We trained and tested our method on datasets from three sources: Medical College of Wisconsin (MCW), Kessler Foundation (KF), and the publicly available Anatomical Tracings of Lesions After Stroke (ATLAS) dataset. Cross-study validation results (with independent training and validation datasets) were obtained to compare with previous methods based on naive Bayes, random forests, and three recently published convolutional neural networks. Model performance was quantified in terms of the Dice coefficient. Training on the KF and MCW images and testing on the ATLAS images yielded a mean Dice coefficient of 0.54. This was reliably better than the next best previous model, UNet, at 0.47. Reversing the train and test datasets yields a mean Dice of 0.47 on KF and MCW images, whereas the next best UNet reaches 0.45. With all three datasets combined, the current system compared to previous methods also attained a reliably higher cross-validation accuracy. It also achieved high Dice values for many smaller lesions that existing methods have difficulty identifying. Overall, our system is a clear improvement over previous methods for automating stroke lesion segmentation, bringing us an important step closer to the inter-rater accuracy level of human experts.
http://arxiv.org/abs/1905.10835
Convolutional neural networks (CNNs) achieve state-of-the-art accuracy in a variety of tasks in computer vision and beyond. One of the major obstacles hindering the ubiquitous use of CNNs for inference on low-power edge devices is their relatively high computational complexity and memory bandwidth requirements. The latter often dominates the energy footprint on modern hardware. In this paper, we introduce a lossy transform coding approach, inspired by image and video compression, designed to reduce the memory bandwidth due to the storage of intermediate activation calculation results. Our method exploits the high correlations between feature maps and adjacent pixels and allows to halve the data transfer volumes to the main memory without re-training. We analyze the performance of our approach on a variety of CNN architectures and demonstrated FPGA implementation of ResNet18 with our approach results in reduction of around 40% in the memory energy footprint compared to quantized network with negligible impact on accuracy. A reference implementation is available at https://github.com/CompressTeam/TransformCodingInference
http://arxiv.org/abs/1905.10830
Face performance capture and reenactment techniques use multiple cameras and sensors, positioned at a distance from the face or mounted on heavy wearable devices. This limits their applications in mobile and outdoor environments. We present EgoFace, a radically new lightweight setup for face performance capture and front-view videorealistic reenactment using a single egocentric RGB camera. Our lightweight setup allows operations in uncontrolled environments, and lends itself to telepresence applications such as video-conferencing from dynamic environments. The input image is projected into a low dimensional latent space of the facial expression parameters. Through careful adversarial training of the parameter-space synthetic rendering, a videorealistic animation is produced. Our problem is challenging as the human visual system is sensitive to the smallest face irregularities that could occur in the final results. This sensitivity is even stronger for video results. Our solution is trained in a pre-processing stage, through a supervised manner without manual annotations. EgoFace captures a wide variety of facial expressions, including mouth movements and asymmetrical expressions. It works under varying illuminations, background, movements, handles people from different ethnicities and can operate in real time.
http://arxiv.org/abs/1905.10822
Algorithms typically come with tunable parameters that have a considerable impact on the computational resources they consume. Too often, practitioners must hand-tune the parameters, a tedious and error-prone task. A recent line of research provides algorithms that return nearly-optimal parameters from within a finite set. These algorithms can be used when the parameter space is infinite by providing as input a random sample of parameters. This data-independent discretization, however, might miss pockets of nearly-optimal parameters: prior research has presented scenarios where the only viable parameters lie within an arbitrarily small region. We provide an algorithm that learns a finite set of promising parameters from within an infinite set. Our algorithm can help compile a configuration portfolio, or it can be used to select the input to a configuration algorithm for finite parameter spaces. Our approach applies to any configuration problem that satisfies a simple yet ubiquitous structure: the algorithm’s performance is a piecewise constant function of its parameters. Prior research has exhibited this structure in domains from integer programming to clustering. For these types of combinatorial problems, this is the first configuration algorithm beyond exhaustive search whose output provably competes with the best parameters from an infinite space.
http://arxiv.org/abs/1905.10819
We describe a shared control methodology that can, without knowledge of the task, be used to improve a human’s control of a dynamic system, be used as a training mechanism, and be used in conjunction with Imitation Learning to generate autonomous policies that recreate novel behaviors. Our algorithm introduces autonomy that assists the human partner by enforcing safety and stability constraints. The autonomous agent has no a priori knowledge of the desired task and therefore only adds control information when there is concern for the safety of the system. We evaluate the efficacy of our approach with a human subjects study consisting of 20 participants. We find that our shared control algorithm significantly improves the rate at which users are able to successfully execute novel behaviors. Experimental results suggest that the benefits of our safety-aware shared control algorithm also extend to the human partner’s understanding of the system and their control skill. Finally, we demonstrate how a combination of our safety-aware shared control algorithm and Imitation Learning can be used to autonomously recreate the demonstrated behaviors.
http://arxiv.org/abs/1905.10814
Spelling error correction is an important problem in natural language processing, as a prerequisite for good performance in downstream tasks as well as an important feature in user-facing applications. For texts in Polish language, there exist works on specific error correction solutions, often developed for dealing with specialized corpora, but not evaluations of many different approaches on big resources of errors. We begin to address this problem by testing some basic and promising methods on PlEWi, a corpus of annotated spelling extracted from Polish Wikipedia. These modules may be further combined with appropriate solutions for error detection and context awareness. Following our results, combining edit distance with cosine distance of semantic vectors may be suggested for interpretable systems, while an LSTM, particularly enhanced by ELMo embeddings, seems to offer the best raw performance.
http://arxiv.org/abs/1905.10810
Word2Vec (W2V) and Glove are popular word embedding algorithms that perform well on a variety of natural language processing tasks. The algorithms are fast, efficient and their embeddings widely used. Moreover, the W2V algorithm has recently been adopted in the field of graph embedding, where it underpins several leading algorithms. However, despite their ubiquity and the relative simplicity of their common architecture, what the embedding parameters of W2V and Glove learn and why that it useful in downstream tasks largely remains a mystery. We show that different interactions of PMI vectors encode semantic properties that can be captured in low dimensional word embeddings by suitable projection, theoretically explaining why the embeddings of W2V and Glove work, and, in turn, revealing an interesting mathematical interconnection between the semantic relationships of relatedness, similarity, paraphrase and analogy.
http://arxiv.org/abs/1805.12164
Different from the traditional classification tasks which assume mutual exclusion of labels, hierarchical multi-label classification (HMLC) aims to assign multiple labels to every instance with the labels organized under hierarchical relations. In fact, linguistic ontologies are intrinsic hierarchies. Besides the labels, the conceptual relations between words can also form hierarchical structures. Thus it can be a challenge to learn mappings from the word space to the label space, and vice versa. We propose to model the word and label hierarchies by embedding them jointly in the hyperbolic space. The main reason is that the tree-likeness of the hyperbolic space matches the complexity of symbolic data with hierarchical structures. A new hyperbolic interaction model (HyperIM) is designed to learn the label-aware document representations and make predictions for HMLC. Extensive experiments are conducted on three benchmark datasets. The results have demonstrated that the new model can realistically capture the complex data structures and further improve the performance for HMLC comparing with the state-of-the-art methods. To facilitate future research, our code is publicly available.
http://arxiv.org/abs/1905.10802
The knowledge base completion problem is the problem of inferring missing information from existing facts in knowledge bases. Path-ranking based methods use sequences of relations as general patterns of paths for prediction. However, these patterns usually lack accuracy because they are generic and can often apply to widely varying scenarios. We leverage type hierarchies of entities to create a new class of path patterns that are both discriminative and generalizable. Then we propose an attention-based RNN model, which can be trained end-to-end, to discover the new path patterns most suitable for the data. Experiments conducted on two benchmark knowledge base completion datasets demonstrate that the proposed model outperforms existing methods by a statistically significant margin. Our quantitative analysis of the path patterns shows that they balance between generalization and discrimination.
http://arxiv.org/abs/1905.10799
Explaining a deep learning model can help users understand its behavior and allow researchers to discern its shortcomings. Recent work has primarily focused on explaining models for tasks like image classification or visual question answering. In this paper, we introduce an explanation approach for image similarity models, where a model’s output is a semantic feature representation rather than a classification. In this task, an explanation depends on both of the input images, so standard methods do not apply. We propose an explanation method that pairs a saliency map identifying important image regions with an attribute that best explains the match. We find that our explanations are more human-interpretable than saliency maps alone, and can also improve performance on the classic task of attribute recognition. The ability of our approach to generalize is demonstrated on two datasets from very different domains, Polyvore Outfits and Animals with Attributes 2.
http://arxiv.org/abs/1905.10797
This work presents an online learning-based control method for improved trajectory tracking of unmanned aerial vehicles using both deep learning and expert knowledge. The proposed method does not require the exact model of the system to be controlled, and it is robust against variations in system dynamics as well as operational uncertainties. The learning is divided into two phases: offline (pre-)training and online (post-)training. In the former, a conventional controller performs a set of trajectories and, based on the input-output dataset, the deep neural network (DNN)-based controller is trained. In the latter, the trained DNN, which mimics the conventional controller, controls the system. Unlike the existing papers in the literature, the network is still being trained for different sets of trajectories which are not used in the training phase of DNN. Thanks to the rule-base, which contains the expert knowledge, the proposed framework learns the system dynamics and operational uncertainties in real-time. The experimental results show that the proposed online learning-based approach gives better trajectory tracking performance when compared to the only offline trained network.
http://arxiv.org/abs/1905.10796
We are interested in learning models of intuitive physics similar to the ones that animals use for navigation, manipulation and planning. In addition to learning general physical principles, however, we are also interested in learning ``on the fly’’, from a few experiences, physical properties specific to new environments. We do all this in an unsupervised manner, using a meta-learning formulation where the goal is to predict videos containing demonstrations of physical phenomena, such as objects moving and colliding with a complex background. We introduce the idea of summarizing past experiences in a very compact manner, in our case using dynamic images, and show that this can be used to solve the problem well and efficiently. Empirically, we show via extensive experiments and ablation studies, that our model learns to perform physical predictions that generalize well in time and space, as well as to a variable number of interacting physical objects.
http://arxiv.org/abs/1905.10793
Ensemble Decision Systems offer a unique form of decision making that allows a collection of algorithms to reason together about a problem. Each individual algorithm has its own inherent strengths and weaknesses, and often it is difficult to overcome the weaknesses while retaining the strengths. Instead of altering the properties of the algorithm, the Ensemble Decision System augments the performance with other algorithms that have complementing strengths. This work outlines different options for building an Ensemble Decision System as well as providing analysis on its performance compared to the individual components of the system with interesting results, showing an increase in the generality of the algorithms without significantly impeding performance.
http://arxiv.org/abs/1905.10792
Face expression recognition is still a complex task, particularly due to the presence of head pose variations. Although face alignment approaches are becoming increasingly accurate for characterizing facial regions, it is important to consider the impact of these approaches when they are used for other related tasks such as head pose registration or facial expression recognition. In this paper, we compare the performance of recent face alignment approaches to highlight the most appropriate techniques for preserving facial geometry when correcting the head pose variation. Also, we highlight the most suitable techniques that locate facial landmarks in the presence of head pose variations and facial expressions.
http://arxiv.org/abs/1905.10784
Fine-grained temporal action parsing is important in many applications, such as daily activity understanding, human motion analysis, surgical robotics and others requiring subtle and precise operations in a long-term period. In this paper we propose a novel bilinear pooling operation, which is used in intermediate layers of a temporal convolutional encoder-decoder net. In contrast to other work, our proposed bilinear pooling is learnable and hence can capture more complex local statistics than the conventional counterpart. In addition, we introduce exact lower-dimension representations of our bilinear forms, so that the dimensionality is reduced with neither information loss nor extra computation. We perform intensive experiments to quantitatively analyze our model and show the superior performances to other state-of-the-art work on various datasets.
http://arxiv.org/abs/1812.01922
Recent deep learning based face recognition methods have achieved great performance, but it still remains challenging to recognize very low-resolution query face like 28x28 pixels when CCTV camera is far from the captured subject. Such face with very low-resolution is totally out of detail information of the face identity compared to normal resolution in a gallery and hard to find corresponding faces therein. To this end, we propose a Resolution Invariant Model (RIM) for addressing such cross-resolution face recognition problems, with three distinct novelties. First, RIM is a novel and unified deep architecture, containing a Face Hallucination sub-Net (FHN) and a Heterogeneous Recognition sub-Net (HRN), which are jointly learned end to end. Second, FHN is a well-designed tri-path Generative Adversarial Network (GAN) which simultaneously perceives facial structure and geometry prior information, i.e. landmark heatmaps and parsing maps, incorporated with an unsupervised cross-domain adversarial training strategy to super-resolve very low-resolution query image to its 8x larger ones without requiring them to be well aligned. Third, HRN is a generic Convolutional Neural Network (CNN) for heterogeneous face recognition with our proposed residual knowledge distillation strategy for learning discriminative yet generalized feature representation. Quantitative and qualitative experiments on several benchmarks demonstrate the superiority of the proposed model over the state-of-the-arts. Codes and models will be released upon acceptance.
http://arxiv.org/abs/1905.10777
Machine learning promises methods that generalize well from finite labeled data. However, the brittleness of existing neural net approaches is revealed by notable failures, such as the existence of adversarial examples that are misclassified despite being nearly identical to a training example, or the inability of recurrent sequence-processing nets to stay on track without teacher forcing. We introduce a method, which we refer to as \emph{state reification}, that involves modeling the distribution of hidden states over the training data and then projecting hidden states observed during testing toward this distribution. Our intuition is that if the network can remain in a familiar manifold of hidden space, subsequent layers of the net should be well trained to respond appropriately. We show that this state-reification method helps neural nets to generalize better, especially when labeled data are sparse, and also helps overcome the challenge of achieving robust generalization with adversarial training.
http://arxiv.org/abs/1905.11382
A testbed has recently been introduced that evolves controllers for arbitrary hover-capable UAVs, with evaluations occurring directly on the robot. To prepare the testbed for real-world deployment, we investigate the effects of state-space limitations brought about by physical tethering (which prevents damage to the UAV during stochastic tuning), on the generality of the evolved controllers. We identify generalisation issues in some controllers, and propose an improved method that comprises two stages: in the first stage, controllers are evolved as normal using standard tethers, but experiments are terminated when the population displays basic flight competency. Optimisation then continues on a much less restrictive tether, effectively free-flying, and is allowed to explore a larger state-space envelope. We compare the two methods on a hover task using a real UAV, and show that more general solutions are generated in fewer generations using the two-stage approach. A secondary experiment undertakes a sensitivity analysis of the evolved controllers.
http://arxiv.org/abs/1905.10762
On-board processing elements on UAVs are currently inadequate for training and inference of Deep Neural Networks. This is largely due to the energy consumption of memory accesses in such a network. HadaNets introduce a flexible train-from-scratch tensor quantization scheme by pairing a full precision tensor to a binary tensor in the form of a Hadamard product. Unlike wider reduced precision neural network models, we preserve the train-time parameter count, thus out-performing XNOR-Nets without a train-time memory penalty. Such training routines could see great utility in semi-supervised online learning tasks. Our method also offers advantages in model compression, as we reduce the model size of ResNet-18 by 7.43 times with respect to a full precision model without utilizing any other compression techniques. We also demonstrate a ‘Hadamard Binary Matrix Multiply’ kernel, which delivers a 10-fold increase in performance over full precision matrix multiplication with a similarly optimized kernel.
http://arxiv.org/abs/1905.10759
Partial domain adaptation (PDA) extends standard domain adaptation to a more realistic scenario where the target domain only has a subset of classes from the source domain. The key challenge of PDA is how to select the relevant samples in the shared classes for knowledge transfer. Previous PDA methods tackle this problem by re-weighting the source samples based on the prediction of classifier or discriminator, thus discarding the pixel-level information. In this paper, to utilize both high-level and pixel-level information, we propose a reinforced transfer network (RTNet), which is the first work to apply reinforcement learning to address the PDA problem. The RTNet simultaneously mitigates the negative transfer by adopting a reinforced data selector to filter out outlier source classes, and promotes the positive transfer by employing a domain adaptation model to minimize the distribution discrepancy in the shared label space. Extensive experiments indicate that RTNet can achieve state-of-the-art performance for partial domain adaptation tasks on several benchmark datasets. Codes and datasets will be available online.
http://arxiv.org/abs/1905.10756
Text infilling is defined as a task for filling in the missing part of a sentence or paragraph, which is suitable for many real-world natural language generation scenarios. However, given a well-trained sequential generative model, generating missing symbols conditioned on the context is challenging for existing greedy approximate inference algorithms. In this paper, we propose an iterative inference algorithm based on gradient search, which is the first inference algorithm that can be broadly applied to any neural sequence generative models for text infilling tasks. We compare the proposed method with strong baselines on three text infilling tasks with various mask ratios and different mask strategies. The results show that our proposed method is effective and efficient for fill-in-the-blank tasks, consistently outperforming all baselines.
http://arxiv.org/abs/1905.10752
We present a model for separating a set of voices out of a sound mixture containing an unknown number of sources. Our Attentional Gating Network (AGN) uses a variable attentional context to specify which speakers in the mixture are of interest. The attentional context is specified by an embedding vector which modifies the processing of a neural network through an additive bias. Individual speaker embeddings are learned to separate a single speaker while superpositions of the individual speaker embeddings are used to separate sets of speakers. We first evaluate AGN on a traditional single speaker separation task and show an improvement of 9% with respect to comparable models. Then, we introduce a new task to separate an arbitrary subset of voices from a mixture of an unknown-sized set of voices, inspired by the human ability to separate a conversation of interest from background chatter at a cafeteria. We show that AGN is the only model capable of solving this task, performing only 7% worse than on the single speaker separation task.
http://arxiv.org/abs/1905.10751
In unsupervised domain adaptation, existing methods utilizing the boundary decision have achieved remarkable performance, but they lack analysis of the relationship between decision boundary and features. In our work, we propose a new principle that adaptive classifiers and transferable features can be obtained in the target domain by learning smooth representations. We analyze the relationship between decision boundary and ambiguous target features in terms of smoothness. Thereafter, local smooth discrepancy is defined to measure the smoothness of a sample and detect sensitive samples which are easily misclassified. To strengthen the smoothness, sensitive samples are corrected in feature space by optimizing local smooth discrepancy. Moreover, the generalization error upper bound is derived theoretically. Finally, We evaluate our method in several standard benchmark datasets. Empirical evidence shows that the proposed method is comparable or superior to the state-of-the-art methods and local smooth discrepancy is a valid metric to evaluate the performance of a domain adaptation method.
http://arxiv.org/abs/1905.10748
Existing methods for AI-generated artworks still struggle with generating high-quality stylized content, where high-level semantics are preserved, or separating fine-grained styles from various artists. We propose a novel Generative Adversarial Disentanglement Network which can fully decompose complex anime illustrations into style and content. Training such model is challenging, since given a style, various content data may exist but not the other way round. In particular, we disentangle two complementary factors of variations, where one of the factors is labelled. Our approach is divided into two stages, one that encodes an input image into a style independent content, and one based on a dual-conditional generator. We demonstrate the ability to generate high-fidelity anime portraits with a fixed content and a large variety of styles from over a thousand artists, and vice versa, using a single end-to-end network and with applications in style transfer. We show this unique capability as well as superior output to the current state-of-the-art.
http://arxiv.org/abs/1905.10742
Machine learning models are vulnerable to adversarial examples. Iterative adversarial training has shown promising results against strong white-box attacks. However, adversarial training is very expensive, and every time a model needs to be protected, such expensive training scheme needs to be performed. In this paper, we propose to apply iterative adversarial training scheme to an external auto-encoder, which once trained can be used to protect other models directly. We empirically show that our model outperforms other purifying-based methods against white-box attacks, and transfers well to directly protect other base models with different architectures.
http://arxiv.org/abs/1905.10729
Evaluating AMR parsing accuracy involves comparing pairs of AMR graphs. The only existing evaluation metric, Smatch (Cai and Knight, 2013), searches for one-to-one mappings between the nodes of two AMRs with a greedy hill-climbing algorithm, which leads to search errors. We propose SemBleu, a robust metric that extends BLEU (Papineni et al., 2002) to AMRs. It does not suffer from search errors and considers non-local correspondences in addition to local ones. SemBleu is fully content-driven and punishes situations where a system output does not preserve most information from the input. Preliminary experiments on both sentence and corpus levels show that SemBleu has slightly higher consistency with human judgments than Smatch. Our code and data at this http URL com/freesunshine0316/sembleu.
http://arxiv.org/abs/1905.10726
There is an immense literature focused on estimating the curvature of an unknown surface from point cloud dataset. Most existing algorithms estimate the curvature indirectly, that is, to estimate the surface locally by some basis functions and then calculate the curvature of such surface as an estimate of the curvature. Recently several methods have been proposed to estimate the curvature directly. However, these algorithms lack of theoretical guarantee on estimation error on small to moderate datasets. In this paper, we propose a direct and efficient method to estimate the curvature for oriented point cloud data without any surface approximation. In fact, we estimate the Weingarten map using a least square method, so that Gaussian curvature, mean curvature and principal curvatures can be obtained automatically from the Weingarten map. We show the convergence rate of our Weingarten Map Estimation (WME) algorithm is $n^{-2/3}$ both theoretically and numerically. Finally, we apply our method to point cloud simplification and surface reconstruction.
http://arxiv.org/abs/1905.10725
Answer selection (answer ranking) is one of the key steps in many kinds of question answering (QA) applications, where deep models have achieved state-of-the-art performance. Among these deep models, recurrent neural network (RNN) based models are most popular, typically with better performance than convolutional neural network (CNN) based models. Nevertheless, it is difficult for RNN based models to capture the information about long-range dependency among words in the sentences of questions and answers. In this paper, we propose a new deep model, called gated group self-attention (GGSA), for answer selection. GGSA is inspired by global self-attention which is originally proposed for machine translation and has not been explored in answer selection. GGSA tackles the problem of global self-attention that local and global information cannot be well distinguished. Furthermore, an interaction mechanism between questions and answers is also proposed to enhance GGSA by a residual structure. Experimental results on two popular QA datasets show that GGSA can outperform existing answer selection models to achieve state-of-the-art performance. Furthermore, GGSA can also achieve higher accuracy than global self-attention for the answer selection task, with a lower computation cost.
http://arxiv.org/abs/1905.10720
Answer selection is an important subtask of question answering (QA), where deep models usually achieve better performance. Most deep models adopt question-answer interaction mechanisms, such as attention, to get vector representations for answers. When these interaction based deep models are deployed for online prediction, the representations of all answers need to be recalculated for each question. This procedure is time-consuming for deep models with complex encoders like BERT which usually have better accuracy than simple encoders. One possible solution is to store the matrix representation (encoder output) of each answer in memory to avoid recalculation. But this will bring large memory cost. In this paper, we propose a novel method, called hashing based answer selection (HAS), to tackle this problem. HAS adopts a hashing strategy to learn a binary matrix representation for each answer, which can dramatically reduce the memory cost for storing the matrix representations of answers. Hence, HAS can adopt complex encoders like BERT in the model, but the online prediction of HAS is still fast with a low memory cost. Experimental results on three popular answer selection datasets show that HAS can outperform existing models to achieve state-of-the-art performance.
http://arxiv.org/abs/1905.10718
Online learning algorithms update models via one sample per iteration, thus efficient to process large-scale datasets and useful to detect malicious events for social benefits, such as disease outbreak and traffic congestion on the fly. However, existing algorithms for graph-structured models focused on the offline setting and the least square loss, incapable for online setting, while methods designed for online setting cannot be directly applied to the problem of complex (usually non-convex) graph-structured sparsity model. To address these limitations, in this paper we propose a new algorithm for graph-structured sparsity constraint problems under online setting, which we call \textsc{GraphDA}. The key part in \textsc{GraphDA} is to project both averaging gradient (in dual space) and primal variables (in primal space) onto lower dimensional subspaces, thus capturing the graph-structured sparsity effectively. Furthermore, the objective functions assumed here are generally convex so as to handle different losses for online learning settings. To the best of our knowledge, \textsc{GraphDA} is the first online learning algorithm for graph-structure constrained optimization problems. To validate our method, we conduct extensive experiments on both benchmark graph and real-world graph datasets. Our experiment results show that, compared to other baseline methods, \textsc{GraphDA} not only improves classification performance, but also successfully captures graph-structured features more effectively, hence stronger interpretability.
http://arxiv.org/abs/1905.10714
Reconstructing 3D shapes from single-view images has been a long-standing research problem and has attracted a lot of attention. In this paper, we present DISN, a Deep Implicit Surface Network that generates a high-quality 3D shape given an input image by predicting the underlying signed distance field. In addition to utilizing global image features, DISN also predicts the local image patch each 3D point sample projects onto and extracts local features from the patch. Combining global and local features significantly improves the accuracy of the predicted signed distance field. To the best of our knowledge, DISN is the first method that constantly captures details such as holes and thin structures present in 3D shapes from single-view images. DISN achieves state-of-the-art single-view reconstruction performance on a variety of shape categories reconstructed from both synthetic and real images. Code is available at github.com/laughtervv/DISN.
http://arxiv.org/abs/1905.10711
Anomaly detection is a problem of great interest in medicine, finance, and other fields where error and fraud need to be detected and corrected. Most deep anomaly detection methods rely on autoencoder reconstruction error. However, we show that this approach has limited value. First, this approach starts to perform poorly when either noise or anomalies contaminate training data, even to a small extent. Second, this approach cannot detect anomalous but simple to reconstruct points. This can be seen even in relatively simple examples, such as feeding a black image to detectors trained on MNIST digits. Here, we introduce a new discriminator-based unsupervised Lipschitz anomaly detector (LAD). We train a Wasserstein discriminator, similar to the ones used in GANs, to detect the difference between the training data and corruptions of the training data. We show that this procedure successfully detects unseen anomalies with guarantees on those that have a certain Wasserstein distance from the data or corrupted training set. Finally, we show results of this system in an electronic medical record dataset of HIV-positive veterans from the veterans aging cohort study (VACS) to establish usability in a medical setting.
http://arxiv.org/abs/1905.10710
Given a sufficiently large training dataset, it is relatively easy to train a modern convolution neural network (CNN) as a required image classifier. However, for the task of fish classification and/or fish detection, if a CNN was trained to detect or classify particular fish species in particular background habitats, the same CNN exhibits much lower accuracy when applied to new/unseen fish species and/or fish habitats. Therefore, in practice, the CNN needs to be continuously fine-tuned to improve its classification accuracy to handle new project-specific fish species or habitats. In this work we present a labelling-efficient method of training a CNN-based fish-detector (the Xception CNN was used as the base) on relatively small numbers (4,000) of project-domain underwater fish/no-fish images from 20 different habitats. Additionally, 17,000 of known negative (that is, missing fish) general-domain (VOC2012) above-water images were used. Two publicly available fish-domain datasets supplied additional 27,000 of above-water and underwater positive/fish images. By using this multi-domain collection of images, the trained Xception-based binary (fish/not-fish) classifier achieved 0.17% false-positives and 0.61% false-negatives on the project’s 20,000 negative and 16,000 positive holdout test images, respectively. The area under the ROC curve (AUC) was 99.94%.
http://arxiv.org/abs/1905.10708
Intelligent agents need a physical understanding of the world to predict the impact of their actions in the future. While learning-based models of the environment dynamics have contributed to significant improvements in sample efficiency compared to model-free reinforcement learning algorithms, they typically fail to generalize to system states beyond the training data, while often grounding their predictions on non-interpretable latent variables. We introduce Interactive Differentiable Simulation (IDS), a differentiable physics engine, that allows for efficient, accurate inference of physical properties of rigid-body systems. Integrated into deep learning architectures, our model is able to accomplish system identification using visual input, leading to an interpretable model of the world whose parameters have physical meaning. We present experiments showing automatic task-based robot design and parameter estimation for nonlinear dynamical systems by automatically calculating gradients in IDS. When integrated into an adaptive model-predictive control algorithm, our approach exhibits orders of magnitude improvements in sample efficiency over model-free reinforcement learning algorithms on challenging nonlinear control domains.
http://arxiv.org/abs/1905.10706
Over the past decade, knowledge graphs became popular for capturing structured domain knowledge. Relational learning models enable the prediction of missing links inside knowledge graphs. More specifically, latent distance approaches model the relationships among entities via a distance between latent representations. Translating embedding models (e.g., TransE) are among the most popular latent distance approaches which use one distance function to learn multiple relation patterns. However, they are not capable of capturing symmetric relations. They also force relations with reflexive patterns to become symmetric and transitive. In order to improve distance based embedding, we propose multi-distance embeddings (MDE). Our solution is based on the idea that by learning independent embedding vectors for each entity and relation one can aggregate contrasting distance functions. Benefiting from MDE, we also develop supplementary distances resolving the above-mentioned limitations of TransE. We further propose an extended loss function for distance based embeddings and show that MDE and TransE are fully expressive using this loss function. Furthermore, we obtain a bound on the size of their embeddings for full expressivity. Our empirical results show that MDE significantly improves the translating embeddings and outperforms several state-of-the-art embedding models on benchmark datasets.
http://arxiv.org/abs/1905.10702
Sudoku is a puzzle well-known to the scientific community with simple rules of completion, which may require a com-plex line of reasoning. This paper addresses the problem of partitioning the Sudoku image into a 1-D array, recognizing digits from the array and representing it as a Constraint Sat-isfaction Problem (CSP). In this paper, we introduce new fea-ture extraction techniques for recognizing digits, which are used with our benchmark classifiers in conjunction with the CSP algorithms to provide performance assessment. Experi-mental results show that application of CSP techniques can decrease the solution’s search time by eliminating incon-sistent values from the search space.
http://arxiv.org/abs/1905.10701
The rise of increasingly more powerful chatbots offers a new way to collect information through conversational surveys, where a chatbot asks open-ended questions, interprets a user’s free-text responses, and probes answers when needed. To investigate the effectiveness and limitations of such a chatbot in conducting surveys, we conducted a field study involving about 600 participants. In this study, half of the participants took a typical online survey on Qualtrics and the other half interacted with an AI-powered chatbot to complete a conversational survey. Our detailed analysis of over 5200 free-text responses revealed that the chatbot drove a significantly higher level of participant engagement and elicited significantly better quality responses in terms of relevance, depth, and readability. Based on our results, we discuss design implications for creating AI-powered chatbots to conduct effective surveys and beyond.
http://arxiv.org/abs/1905.10700
Transferring knowledge from one neural network to another has been shown to be helpful for learning tasks with few training examples. Prevailing fine-tuning methods could potentially contaminate pre-trained features by comparably high energy random noise. This noise is mainly delivered from a careless replacement of task-specific parameters. We analyze theoretically such knowledge contamination for classification tasks and propose a practical and easy to apply method to trap and minimize the contaminant. In our approach, the entropy of the output estimates gets maximized initially and the first back-propagated error is stalled at the output of the last layer. Our proposed method not only outperforms the traditional fine-tuning, but also significantly speeds up the convergence of the learner. It is robust to randomness and independent of the choice of architecture. Overall, our experiments show that the power of transfer learning has been substantially underestimated so far.
http://arxiv.org/abs/1905.10698
Deep Neural Networks (DNNs) are vulnerable to adversarial attacks, especially white-box targeted attacks. One scheme of learning attacks is to design a proper adversarial objective function that leads to the imperceptible perturbation for any test image (e.g., the Carlini-Wagner (C&W) method). Most methods address targeted attacks in the Top-1 manner. In this paper, we propose to learn ordered Top-k attacks (k>= 1) for image classification tasks, that is to enforce the Top-k predicted labels of an adversarial example to be the k (randomly) selected and ordered labels (the ground-truth label is exclusive). To this end, we present an adversarial distillation framework: First, we compute an adversarial probability distribution for any given ordered Top-k targeted labels with respect to the ground-truth of a test image. Then, we learn adversarial examples by minimizing the Kullback-Leibler (KL) divergence together with the perturbation energy penalty, similar in spirit to the network distillation method. We explore how to leverage label semantic similarities in computing the targeted distributions, leading to knowledge-oriented attacks. In experiments, we thoroughly test Top-1 and Top-5 attacks in the ImageNet-1000 validation dataset using two popular DNNs trained with clean ImageNet-1000 train dataset, ResNet-50 and DenseNet-121. For both models, our proposed adversarial distillation approach outperforms the C&W method in the Top-1 setting, as well as other baseline methods. Our approach shows significant improvement in the Top-5 setting against a strong modified C&W method.
http://arxiv.org/abs/1905.10695
Simple cells in primary visual cortex (V1) can be approximated by Gabor filters, and adjacent simple cells tend to have quadrature phase relationship. This paper entertains the hypothesis that a key purpose of such simple cells is to perceive local motions, i.e., displacements of pixels, caused by the relative motions between the agent and the surrounding environment. Specifically, we propose a representational model that couples the vector representations of local image contents with the matrix representations of local pixel displacements. When the image changes from one time frame to the next due to pixel displacements, the vector at each pixel is rotated by a matrix that represents the displacement of this pixel. We show that by learning from pair of images that are deformed versions of each other, we can learn both vector and matrix representations. The units in the learned vector representations reproduce properties of V1 simple cells. The learned model enables perceptual inference of local motions.
http://arxiv.org/abs/1902.03871
This paper presents a conceptually simple and effective Deep Audio-Visual Eembedding for dynamic saliency prediction dubbed ``DAVE”. Several behavioral studies have shown a strong relation between auditory and visual cues for guiding gaze during scene free viewing. The existing video saliency models, however, only consider visual cues for predicting saliency over videos and neglect the auditory information that is ubiquitous in dynamic scenes. We propose a multimodal saliency model that utilizes audio and visual information for predicting saliency in videos. Our model consists of a two-stream encoder and a decoder. First, auditory and visual information are mapped into a feature space using 3D Convolutional Neural Networks (3D CNNs). Then, a decoder combines the features and maps them to a final saliency map. To train such model, data from various eye tracking datasets containing video and audio are pulled together. We further categorised videos into social', nature’, and `miscellaneous’ classes to analyze the models over different content types. Several analyses show that our audio-visual model outperforms video-based models significantly over all scores; overall and over individual categories. Contextual analysis of the model performance over the location of sound source reveals that the audio-visual model behaves similar to humans in attending to the location of sound source. Our endeavour demonstrates that audio is an important signal that can boost video saliency prediction and help getting closer to human performance.
http://arxiv.org/abs/1905.10693
To improve the speed and accuracy of the trace based policy evaluation method TD({\lambda}), under appropriate assumptions, we derive and propose an off-policy compatible method of meta-learning state-based {\lambda}’s online with efficient incremental updates. Furthermore, we prove the derived bias-variance tradeoff minimization method, with slight adjustments, is equivalent to minimizing the overall target error in terms of state based {\lambda}’s. In experiments, the method shows significantly better performance when compared to the existing method and the baselines.
http://arxiv.org/abs/1904.11439
In this paper, we introduce Hierarchical Invertible Neural Transport (HINT), an algorithm that merges Invertible Neural Networks and optimal transport to sample from a posterior distribution in a Bayesian framework. This method exploits a hierarchical architecture to construct a Knothe-Rosenblatt transport map between an arbitrary density and the joint density of hidden variables and observations. After training the map, samples from the posterior can be immediately recovered for any contingent observation. Any underlying model evaluation can be performed fully offline from training without the need of a model-gradient. Furthermore, no analytical evaluation of the prior is necessary, which makes HINT an ideal candidate for sequential Bayesian inference. We demonstrate the efficacy of HINT on two numerical experiments.
http://arxiv.org/abs/1905.10687
Deep-learning based classification algorithms have been shown to be susceptible to adversarial attacks: minor changes to the input of classifiers can dramatically change their outputs, while being imperceptible to humans. In this paper, we present a simple hypothesis about a feature compression property of artificial intelligence (AI) classifiers and present theoretical arguments to show that this hypothesis successfully accounts for the observed fragility of AI classifiers to small adversarial perturbations. Drawing on ideas from information and coding theory, we propose a general class of defenses for detecting classifier errors caused by abnormally small input perturbations. We further show theoretical guarantees for the performance of this detection method. We present experimental results with (a) a voice recognition system, and (b) a digit recognition system using the MNIST database, to demonstrate the effectiveness of the proposed defense methods. The ideas in this paper are motivated by a simple analogy between AI classifiers and the standard Shannon model of a communication system.
https://arxiv.org/abs/1905.11381
Composition of elementary skills into complex behaviors to solve challenging problems is one of the key elements toward building intelligent machines. To date, there has been plenty of work on learning new policies or skills but almost no focus on composing them to perform complex decision-making. In this paper, we propose a policy ensemble composition framework that takes the robot’s primitive policies and learns to compose them concurrently or sequentially through reinforcement learning. We evaluate our method in problems where traditional approaches either fail or exhibit high sample complexity to find a solution. We show that our method not only solves the problems that require both task and motion planning but also exhibits high data efficiency, which is currently one of the main limitations of reinforcement learning.
http://arxiv.org/abs/1905.10681
The current state-of-the-art object recognition algorithms, deep convolutional neural networks (DCNNs), are inspired by the architecture of the mammalian visual system [8], and capable of human-level performance on many tasks [15]. However, even these algorithms make errors. As DCNNs improve at object recognition tasks, they develop representations in their hidden layers that become more similar to those observed in the mammalian brains [24]. This led us to hypothesize that teaching DCNNs to achieve even more brain-like representations could improve their performance. To test this, we trained DCNNs on a composite task, wherein networks were trained to: a) classify images of objects; while b) having intermediate representations that resemble those observed in neural recordings from monkey visual cortex. Compared with DCNNs trained purely for object categorization, DCNNs trained on the composite task had better object recognition performance. Our results outline a new way to regularize object recognition networks, using transfer learning strategies in which the brain serves as a teacher for training DCNNs.
http://arxiv.org/abs/1905.10679
Metric learning has become an attractive field for research on the latest years. Loss functions like contrastive loss, triplet loss or multi-class N-pair loss have made possible generating models capable of tackling complex scenarios with the presence of many classes and scarcity on the number of images per class not only work to build classifiers, but to many other applications where measuring similarity is the key. Deep Neural Networks trained via metric learning also offer the possibility to solve few-shot learning problems. Currently used state of the art loss functions such as triplet and contrastive loss functions, still suffer from slow convergence due to the selection of effective training samples that has been partially solved by the multi-class N-pair loss by simultaneously adding additional samples from the different classes. In this work, we extend triplet and multiclass-N-pair loss function by proposing the constellation loss metric where the distances among all class combinations are simultaneously learned. We have compared our constellation loss for visual class embedding showing that our loss function over-performs the other methods by obtaining more compact clusters while achieving better classification results.
http://arxiv.org/abs/1905.10675
Learning high-quality node embeddings is a key building block for machine learning models that operate on graph data, such as social networks and recommender systems. However, existing graph embedding techniques are unable to cope with fairness constraints, e.g., ensuring that the learned representations do not correlate with certain attributes, such as age or gender. Here, we introduce an adversarial framework to enforce fairness constraints on graph embeddings. Our approach is {\em compositional}—meaning that it can flexibly accommodate different combinations of fairness constraints during inference. For instance, in the context of social recommendations, our framework would allow one user to request that their recommendations are invariant to both their age and gender, while also allowing another user to request invariance to just their age. Experiments on standard knowledge graph and recommender system benchmarks highlight the utility of our proposed framework.
http://arxiv.org/abs/1905.10674
In order to be useful in the real world, AI agents need to plan and act in the presence of others, who may include adversarial and cooperative entities. In this paper, we consider the problem where an autonomous agent needs to act in a manner that clarifies its objectives to cooperative entities while preventing adversarial entities from inferring those objectives. We show that this problem is solvable when cooperative entities and adversarial entities use different types of sensors and/or prior knowledge. We develop two new solution approaches for computing such plans. One approach provides an optimal solution to the problem by using an IP solver to provide maximum obfuscation for adversarial entities while providing maximum legibility for cooperative entities in the environment, whereas the other approach provides a satisficing solution using heuristic-guided forward search to achieve preset levels of obfuscation and legibility for adversarial and cooperative entities respectively. We show the feasibility and utility of our algorithms through extensive empirical evaluation on problems derived from planning benchmarks.
http://arxiv.org/abs/1905.10672