Consider end-to-end training of a multi-modal vs. a single-modal network on a task with multiple input modalities: the multi-modal network receives more information, so it should match or outperform its single-modal counterpart. In our experiments, however, we observe the opposite: the best single-modal network always outperforms the multi-modal network. This observation is consistent across different combinations of modalities and on different tasks and benchmarks. This paper identifies two main causes for this performance drop: first, multi-modal networks are often prone to overfitting due to increased capacity. Second, different modalities overfit and generalize at different rates, so training them jointly with a single optimization strategy is sub-optimal. We address these two problems with a technique we call Gradient Blending, which computes an optimal blend of modalities based on their overfitting behavior. We demonstrate that Gradient Blending outperforms widely-used baselines for avoiding overfitting and achieves state-of-the-art accuracy on various tasks including fine-grained sport classification, human action recognition, and acoustic event detection.
http://arxiv.org/abs/1905.12681
We consider the problem of sparsity-constrained $M$-estimation when both explanatory and response variables have heavy tails (bounded 4-th moments), or a fraction of arbitrary corruptions. We focus on the $k$-sparse, high-dimensional regime where the number of variables $d$ and the sample size $n$ are related through $n \sim k \log d$. We define a natural condition we call the Robust Descent Condition (RDC), and show that if a gradient estimator satisfies the RDC, then Robust Hard Thresholding (IHT using this gradient estimator), is guaranteed to obtain good statistical rates. The contribution of this paper is in showing that this RDC is a flexible enough concept to recover known results, and obtain new robustness results. Specifically, new results include: (a) For $k$-sparse high-dimensional linear- and logistic-regression with heavy tail (bounded 4-th moment) explanatory and response variables, a linear-time-computable median-of-means gradient estimator satisfies the RDC, and hence Robust Hard Thresholding is minimax optimal; (b) When instead of heavy tails we have $O(1/\sqrt{k}\log(nd))$-fraction of arbitrary corruptions in explanatory and response variables, a near linear-time computable trimmed gradient estimator satisfies the RDC, and hence Robust Hard Thresholding is minimax optimal. We demonstrate the effectiveness of our approach in sparse linear, logistic regression, and sparse precision matrix estimation on synthetic and real-world US equities data.
http://arxiv.org/abs/1901.08237
{Radio Frequency Identification technology has gained popularity for cheap and easy deployment. In the realm of manufacturing shopfloor, it can be used to track the location of manufacturing objects to achieve better efficiency. The underlying challenge of localization lies in the non-stationary characteristics of manufacturing shopfloor which calls for an adaptive life-long learning strategy in order to arrive at accurate localization results. This paper presents an evolving model based on a novel evolving intelligent system, namely evolving Type-2 Quantum Fuzzy Neural Network (eT2QFNN), which features an interval type-2 quantum fuzzy set with uncertain jump positions. The quantum fuzzy set possesses a graded membership degree which enables better identification of overlaps between classes. The eT2QFNN works fully in the evolving mode where all parameters including the number of rules are automatically adjusted and generated on the fly. The parameter adjustment scenario relies on decoupled extended Kalman filter method. Our numerical study shows that eT2QFNN is able to deliver comparable accuracy compared to state-of-the-art algorithms.
http://arxiv.org/abs/1805.07715
Grogan et al [11,12] have recently proposed a solution to colour transfer by minimising the Euclidean distance L2 between two probability density functions capturing the colour distributions of two images (palette and target). It was shown to be very competitive to alternative solutions based on Optimal Transport for colour transfer. We show that in fact Grogan et al’s formulation can also be understood as a new robust Optimal Transport based framework with entropy regularisation over marginals.
http://arxiv.org/abs/1905.12678
Classical non-neural dependency parsers put considerable effort on the design of feature functions. Especially, they benefit from information coming from structural features, such as features drawn from neighboring tokens in the dependency tree. In contrast, their BiLSTM-based successors achieve state-of-the-art performance without explicit information about the structural context. In this paper we aim to answer the question: How much structural context are the BiLSTM representations able to capture implicitly? We show that features drawn from partial subtrees become redundant when the BiLSTMs are used. We provide a deep insight into information flow in transition- and graph-based neural architectures to demonstrate where the implicit information comes from when the parsers make their decisions. Finally, with model ablations we demonstrate that the structural context is not only present in the models, but it significantly influences their performance.
http://arxiv.org/abs/1905.12676
Several structure learning algorithms have been proposed towards discovering causal or Bayesian Network (BN) graphs, which is a particularly challenging problem in AI. The performance of these algorithms is evaluated based on the relationship the learned graph has with respect to the ground truth graph. However, there is no agreed scoring function to determine this relationship. Moreover, this paper shows that the commonly used metrics tend to be biased in favour of graphs that minimise the number of edges. The evaluation bias is inconsistent and may lead to evaluating graphs with no edges as superior to graphs with varying numbers of correct and incorrect edges; implying that graphs that minimise edges are often favoured over more complex graphs due to bias rather than overall accuracy. While graphs that are less complex are often desirable, the current metrics encourage algorithms to optimise for simplicity, and to discover graphs with a limited number of edges that do not enable full propagation of evidence. This paper proposes a Balanced Scoring Function (BSF) that eliminates this bias by adjusting the reward function based on the difficulty of discovering an edge, or no edge, proportional to their occurrence rate in the ground truth graph. The BSF score can be used in conjunction with other traditional metrics to provide an alternative and unbiased assessment about the capability of structure learning algorithms in discovering causal or BN graphs.
http://arxiv.org/abs/1905.12666
We introduce a novel framework to build a model that can learn how to segment objects from a collection of images without any human annotation. Our method builds on the observation that the location of object segments can be perturbed locally relative to a given background without affecting the realism of a scene. Our approach is to first train a generative model of a layered scene. The layered representation consists of a background image, a foreground image and the mask of the foreground. A composite image is then obtained by overlaying the masked foreground image onto the background. The generative model is trained in an adversarial fashion against a discriminator, which forces the generative model to produce realistic composite images. To force the generator to learn a representation where the foreground layer corresponds to an object, we perturb the output of the generative model by introducing a random shift of both the foreground image and mask relative to the background. Because the generator is unaware of the shift before computing its output, it must produce layered representations that are realistic for any such random perturbation. Finally, we learn to segment an image by defining an autoencoder consisting of an encoder, which we train, and the pre-trained generator as the decoder, which we freeze. The encoder maps an image to a feature vector, which is fed as input to the generator to give a composite image matching the original input image. Because the generator outputs an explicit layered representation of the scene, the encoder learns to detect and segment objects. We demonstrate this framework on real images of several object categories.
http://arxiv.org/abs/1905.12663
A significant amount of information in today’s world is stored in structured and semi-structured knowledge bases. Efficient and simple methods to query them are essential and must not be restricted to only those who have expertise in formal query languages. The field of semantic parsing deals with converting natural language utterances to logical forms that can be easily executed on a knowledge base. In this survey, we examine the various components of a semantic parsing system and discuss prominent work ranging from the initial rule based methods to the current neural approaches to program synthesis. We also discuss methods that operate using varying levels of supervision and highlight the key challenges involved in the learning of such systems.
http://arxiv.org/abs/1812.00978
Understanding generalization in reinforcement learning (RL) is a significant challenge, as many common assumptions of traditional supervised learning theory do not apply. We focus on the special class of reparameterizable RL problems, where the trajectory distribution can be decomposed using the reparametrization trick. For this problem class, estimating the expected return is efficient and the trajectory can be computed deterministically given peripheral random variables, which enables us to study reparametrizable RL using supervised learning and transfer learning theory. Through these relationships, we derive guarantees on the gap between the expected and empirical return for both intrinsic and external errors, based on Rademacher complexity as well as the PAC-Bayes bound. Our bound suggests the generalization capability of reparameterizable RL is related to multiple factors including “smoothness” of the environment transition, reward and agent policy function class. We also empirically verify the relationship between the generalization gap and these factors through simulations.
http://arxiv.org/abs/1905.12654
Recent progress in natural language generation has raised dual-use concerns. While applications like summarization and translation are positive, the underlying technology also might enable adversaries to generate neural fake news: targeted propaganda that closely mimics the style of real news. Modern computer security relies on careful threat modeling: identifying potential threats and vulnerabilities from an adversary’s point of view, and exploring potential mitigations to these threats. Likewise, developing robust defenses against neural fake news requires us first to carefully investigate and characterize the risks of these models. We thus present a model for controllable text generation called Grover. Given a headline like `Link Found Between Vaccines and Autism,’ Grover can generate the rest of the article; humans find these generations to be more trustworthy than human-written disinformation. Developing robust verification techniques against generators like Grover is critical. We find that best current discriminators can classify neural fake news from real, human-written, news with 73% accuracy, assuming access to a moderate level of training data. Counterintuitively, the best defense against Grover turns out to be Grover itself, with 92% accuracy, demonstrating the importance of public release of strong generators. We investigate these results further, showing that exposure bias – and sampling strategies that alleviate its effects – both leave artifacts that similar discriminators can pick up on. We conclude by discussing ethical issues regarding the technology, and plan to release Grover publicly, helping pave the way for better detection of neural fake news.
http://arxiv.org/abs/1905.12616
We revisit the stochastic variance-reduced policy gradient (SVRPG) method proposed by Papini et al. (2018) for reinforcement learning. We provide an improved convergence analysis of SVRPG and show that it can find an $\epsilon$-approximate stationary point of the performance function within $O(1/\epsilon^{5/3})$ trajectories. This sample complexity improves upon the best known result $O(1/\epsilon^2)$ by a factor of $O(1/\epsilon^{1/3})$. At the core of our analysis is (i) a tighter upper bound for the variance of importance sampling weights, where we prove that the variance can be controlled by the parameter distance between different policies; and (ii) a fine-grained analysis of the epoch length and batch size parameters such that we can significantly reduce the number of trajectories required in each iteration of SVRPG. We also empirically demonstrate the effectiveness of our theoretical claims of batch sizes on reinforcement learning benchmark tasks.
http://arxiv.org/abs/1905.12615
A century ago, discoveries of a serious kind of logical error made separately by several leading mathematicians led to acceptance of a sharply enhanced standard for rigor within what ultimately became the foundation for Computer Science. By 1931, Godel had obtained a definitive and remarkable result: an inherent limitation to that foundation. The resulting limitation is not applicable to actual human cognition, to even the smallest extent, unless both of these extremely brittle assumptions hold: humans are infallible reasoners and reason solely via formal inference rules. Both assumptions are contradicted by empirical data from well-known Cognitive Science experiments. This article investigates how a novel multi-part methodology recasts computability theory within Computer Science to obtain a definitive limitation whose application to human cognition avoids assumptions contradicting empirical data. The limitation applies to individual humans, to finite sets of humans, and more generally to any real-world entity.
http://arxiv.org/abs/1905.13010
Hierarchies are an effective way to boost sample efficiency in reinforcement learning, and computational efficiency in classical planning. However, acquiring hierarchies via hand-design (as in classical planning) is suboptimal, while acquiring them via end-to-end reward based training (as in reinforcement learning) is unstable and still prohibitively expensive. In this paper, we pursue an alternate paradigm for acquiring such hierarchical abstractions (or visuo-motor subroutines), via use of passive first person observation data. We use an inverse model trained on small amounts of interaction data to pseudo-label the passive first person videos with agent actions. Visuo-motor subroutines are acquired from these pseudo-labeled videos by learning a latent intent-conditioned policy that predicts the inferred pseudo-actions from the corresponding image observations. We demonstrate our proposed approach in context of navigation, and show that we can successfully learn consistent and diverse visuo-motor subroutines from passive first-person videos. We demonstrate the utility of our acquired visuo-motor subroutines by using them as is for exploration, and as sub-policies in a hierarchical RL framework for reaching point goals and semantic goals. We also demonstrate behavior of our subroutines in the real world, by deploying them on a real robotic platform. Project website with videos, code and data: https://ashishkumar1993.github.io/subroutines/.
http://arxiv.org/abs/1905.12612
We prove bounds on the generalization error of convolutional networks. The bounds are in terms of the training loss, the number of parameters, the Lipschitz constant of the loss and the distance from the weights to the initial weights. They are independent of the number of pixels in the input, and the height and width of hidden feature maps; to our knowledge, they are the first bounds for deep convolutional networks with this property. We present experiments with CIFAR-10 and a scaled-down variant, along with varying hyperparameters of a deep convolutional network, comparing our bounds with practical generalization gaps.
http://arxiv.org/abs/1905.12600
Word sense induction (WSI) is the task of unsupervised clustering of word usages within a sentence to distinguish senses. Recent work obtain strong results by clustering lexical substitutes derived from pre-trained RNN language models (ELMo). Adapting the method to BERT improves the scores even further. We extend the previous method to support a dynamic rather than a fixed number of clusters as supported by other prominent methods, and propose a method for interpreting the resulting clusters by associating them with their most informative substitutes. We then perform extensive error analysis revealing the remaining sources of errors in the WSI task. Our code is available at https://github.com/asafamr/bertwsi.
http://arxiv.org/abs/1905.12598
In recent years, several automatic segmentation methods have been proposed for blood vessels in retinal fundus images, ranging from using cheap and fast trainable filters to complicated neural networks and even deep learning. One example of a filted-based segmentation method is B-COSFIRE. In this approach the image filter is trained with example prototype patterns, to which the filter becomes selective by finding points in a Difference of Gaussian response on circles around the center with large intensity variation. In this paper we discuss and evaluate several of these vessel segmentation methods. We take a closer look at B-COSFIRE and study the performance of B-COSFIRE on the recently published IOSTAR dataset by experiments and we examine how the parameter values affect the performance. In the experiment we manage to reach a segmentation accuracy of 0.9419. Based on our findings we discuss when B-COSFIRE is the preferred method to use and in which circumstances it could be beneficial to use a more (computationally) complex segmentation method. We also shortly discuss areas beyond blood vessel segmentation where these methods can be used to segment elongated structures, such as rivers in satellite images or nerves of a leaf.
http://arxiv.org/abs/1905.12596
The incorporation of macro-actions (temporally extended actions) into multi-agent decision problems has the potential to address the curse of dimensionality associated with such decision problems. Since macro-actions last for stochastic durations, multiple agents executing decentralized policies in cooperative environments must act asynchronously. We present an algorithm that modifies generalized advantage estimation for temporally extended actions, allowing a state-of-the-art policy optimization algorithm to optimize policies in Dec-POMDPs in which agents act asynchronously. We show that our algorithm is capable of learning optimal policies in two cooperative domains, one involving real-time bus holding control and one involving wildfire fighting with unmanned aircraft. Our algorithm works by framing problems as “event-driven decision processes,” which are scenarios in which the sequence and timing of actions and events are random and governed by an underlying stochastic process. In addition to optimizing policies with continuous state and action spaces, our algorithm also facilitates the use of event-driven simulators, which do not require time to be discretized into time-steps. We demonstrate the benefit of using event-driven simulation in the context of multiple agents taking asynchronous actions. We show that fixed time-step simulation risks obfuscating the sequence in which closely separated events occur, adversely affecting the policies learned. In addition, we show that arbitrarily shrinking the time-step scales poorly with the number of agents.
http://arxiv.org/abs/1709.06656
E-commerce web applications are almost ubiquitous in our day to day life, however as useful as they are, most of them have little to no adaptation to user needs, which in turn can cause both lower conversion rates as well as unsatisfied customers. We propose a machine learning system which learns the user behaviour from multiple previous sessions and predicts useful metrics for the current session. In turn, these metrics can be used by the applications to customize and better target the customer, which can mean anything from offering better offers of specific products, targeted notifications or placing smart ads. The data used for the learning algorithm is extracted from Google Analytics Enhanced E-commerce, which is enabled by most e-commerce websites and thus the system can be used by any such merchant. In order to learn the user patterns, only its behaviour features were used, which don’t include names, gender or any other personal information that could identify the user. The learning model that was used is a double recurrent neural network which learns both intra-session and inter-session features. The model predicts for each session a probability score for each of the defined target classes.
http://arxiv.org/abs/1905.12595
A continual learning agent should be able to build on top of existing knowledge to learn on new data quickly while minimizing forgetting. Current intelligent systems based on neural network function approximators arguably do the opposite—they are highly prone to forgetting and rarely trained to facilitate future learning. One reason for this poor behavior is that they learn from a representation that is not explicitly trained for these two goals. In this paper, we propose MRCL, an objective to explicitly learn representations that accelerate future learning and are robust to forgetting under online updates in continual learning. The idea is to optimize the representation such that online updates minimize error on all samples with little forgetting. We show that it is possible to learn representations that are more effective for online updating and that sparsity naturally emerges in these representations. Moreover, our method is complementary to existing continual learning strategies, like MER, which can learn more effectively from representations learned by our objective. Finally, we demonstrate that a basic online updating strategy with our learned representation is competitive with rehearsal based methods for continual learning. We release an implementation of our method at https://github.com/khurramjaved96/mrcl .
http://arxiv.org/abs/1905.12588
Blind image deblurring remains a topic of enduring interest. Learning based approaches, especially those that employ neural networks have emerged to complement traditional model based methods and in many cases achieve vastly enhanced performance. That said, neural network approaches are generally empirically designed and the underlying structures are difficult to interpret. In recent years, a promising technique called algorithm unrolling has been developed that has helped connect iterative algorithms such as those for sparse coding to neural network architectures. However, such connections have not been made yet for blind image deblurring. In this paper, we propose a neural network architecture based on this idea. We first present an iterative algorithm that may be considered as a generalization of the traditional total-variation regularization method in the gradient domain. We then unroll the algorithm to construct a neural network for image deblurring which we refer to as Deep Unrolling for Blind Deblurring (DUBLID). Key algorithm parameters are learned with the help of training images. Our proposed deep network DUBLID achieves significant practical performance gains while enjoying interpretability at the same time. Extensive experimental results show that DUBLID outperforms many state-of-the-art methods and in addition is computationally faster.
http://arxiv.org/abs/1902.03493
Background: Finding biomedical named entities is one of the most essential tasks in biomedical text mining. Recently, deep learning-based approaches have been applied to biomedical named entity recognition (BioNER) and showed promising results. However, as deep learning approaches need an abundant amount of training data, a lack of data can hinder performance. BioNER datasets are scarce resources and each dataset covers only a small subset of entity types. Furthermore, many bio entities are polysemous, which is one of the major obstacles in named entity recognition. Results: To address the lack of data and the entity type misclassification problem, we propose CollaboNet which utilizes a combination of multiple NER models. In CollaboNet, models trained on a different dataset are connected to each other so that a target model obtains information from other collaborator models to reduce false positives. Every model is an expert on their target entity type and takes turns serving as a target and a collaborator model during training time. The experimental results show that CollaboNet can be used to greatly reduce the number of false positives and misclassified entities including polysemous words. CollaboNet achieved state-of-the-art performance in terms of precision, recall and F1 score. Conclusions: We demonstrated the benefits of combining multiple models for BioNER. Our model has successfully reduced the number of misclassified entities and improved the performance by leveraging multiple datasets annotated for different entity types. Given the state-of-the-art performance of our model, we believe that CollaboNet can improve the accuracy of downstream biomedical text mining applications such as bio-entity relation extraction.
http://arxiv.org/abs/1809.07950
This paper proposes a new approach, Flat2Layout, for estimating general indoor room layout from a single-view RGB image whereas existing methods can only produce layout topologies captured from the box-shaped room. The proposed flat representation encodes the layout information into row vectors which are treated as the training target of the deep model. A dynamic programming based postprocessing is employed to decode the estimated flat output from the deep model into the final room layout. Flat2Layout achieves state-of-the-art performance on existing room layout benchmark. This paper also constructs a benchmark for validating the performance on general layout topologies, where Flat2Layout achieves good performance on general room types. Flat2Layout is applicable on more scenario for layout estimation and would have an impact on applications of Scene Modeling, Robotics, and Augmented Reality.
http://arxiv.org/abs/1905.12571
Water quality is of great importance for humans and for the environment and has to be monitored continuously. It is determinable through proxies such as the chlorophyll $a$ concentration, which can be monitored by remote sensing techniques. This study focuses on the trade-off between the spatial and the spectral resolution of six simulated satellite-based data sets when estimating the chlorophyll $a$ concentration with supervised machine learning models. The initial dataset for the spectral simulation of the satellite missions contains spectrometer data and measured chlorophyll $a$ concentration of 13 different inland waters. Focusing on the regression performance, it appears that the machine learning models achieve almost as good results with the simulated Sentinel data as with the simulated hyperspectral data. Regarding the applicability, the Sentinel 2 mission is the best choice for small inland waters due to its high spatial and temporal resolution in combination with a suitable spectral resolution.
http://arxiv.org/abs/1905.12563
Despite renewed interest in emergent language simulations with neural networks, little is known about the basic properties of the induced code, and how they compare to human language. One fundamental characteristic of the latter, known as Zipf’s Law of Abbreviation (ZLA), is that more frequent words are efficiently associated to shorter strings. We study whether the same pattern emerges when two neural networks, a “speaker” and a “listener”, are trained to play a signaling game. Surprisingly, we find that networks develop an \emph{anti-efficient} encoding scheme, in which the most frequent inputs are associated to the longest messages, and messages in general are skewed towards the maximum length threshold. This anti-efficient code appears easier to discriminate for the listener, and, unlike in human communication, the speaker does not impose a contrasting least-effort pressure towards brevity. Indeed, when the cost function includes a penalty for longer messages, the resulting message distribution starts respecting ZLA. Our analysis stresses the importance of studying the basic features of emergent communication in a highly controlled setup, to ensure the latter will not strand too far from human language. Moreover, we present a concrete illustration of how different functional pressures can lead to successful communication codes that lack basic properties of human language, thus highlighting the role such pressures play in the latter.
http://arxiv.org/abs/1905.12561
This paper presents an approach for estimating the operational range for mobile robot exploration on a single battery discharge. Deploying robots in the wild usually requires uninterrupted energy sources to maintain the robot’s mobility throughout the entire mission. However, for most endeavors into the unknown environments, recharging is usually not an option, due to the lack of pre-installed recharging stations or other mission constraints. In these cases, the ability to model the on-board energy consumption and estimate the operational range is crucial to prevent running out of battery in the wild. To this end, this work describes our recent findings that quantitatively break down the robot’s on-board energy consumption and predict the operational range to guarantee safe mission completion on a single battery discharge cycle. Two range estimators with different levels of generality and model fidelity are presented, whose performances were validated on physical robot platforms in both indoor and outdoor environments. Model performance metrics are also presented as benchmarks.
http://arxiv.org/abs/1905.12559
We propose a new type of representation learning method that models words, phrases and sentences seamlessly. Our method does not depend on word segmentation and any human-annotated resources (e.g., word dictionaries), yet it is very effective for noisy corpora written in unsegmented languages such as Chinese and Japanese. The main idea of our method is to ignore word boundaries completely (i.e., segmentation-free), and construct representations for all character $n$-grams in a raw corpus with embeddings of compositional sub-$n$-grams. Although the idea is simple, our experiments on various benchmarks and real-world datasets show the efficacy of our proposal.
http://arxiv.org/abs/1809.00918
Generating coherent and cohesive long-form texts is a challenging task. Previous works relied on large amounts of human-generated texts to train neural language models. However, few attempted to explicitly improve neural language models from the perspectives of coherence and cohesion. In this work, we propose a new neural language model that is equipped with two neural discriminators which provide feedback signals at the levels of sentence (cohesion) and paragraph (coherence). Our model is trained using a simple yet efficient variant of policy gradient, called negative-critical sequence training, which is proposed to eliminate the need of training a separate critic for estimating baseline. Results demonstrate the effectiveness of our approach, showing improvements over the strong baseline – recurrent attention-based bidirectional MLE-trained neural language model.
http://arxiv.org/abs/1811.00511
The Wahba problem, also known as rotation search, seeks to find the best rotation to align two sets of vector observations given putative correspondences, and is a fundamental routine in many computer vision and robotics applications. This work proposes the first polynomial-time certifiably optimal approach for solving the Wahba problem when a large number of vector observations are outliers. Our first contribution is to formulate the Wahba problem using a Truncated Least Squares (TLS) cost that is insensitive to a large fraction of spurious correspondences. The second contribution is to rewrite the problem using unit quaternions and show that the TLS cost can be framed as a Quadratically-Constrained Quadratic Program (QCQP). Since the resulting optimization is still highly non-convex and hard to solve globally, our third contribution is to develop a convex Semidefinite Programming (SDP) relaxation. We show that while a naive relaxation performs poorly in general, our relaxation is tight even in the presence of large noise and outliers. We validate the proposed algorithm, named QUASAR (QUAternion-based Semidefinite relAxation for Robust alignment), in both synthetic and real datasets showing that the algorithm outperforms RANSAC, robust local optimization techniques, and global outlier-removal methods. QUASAR is able to compute certifiably optimal solutions (i.e. the relaxation is exact) even in the case when 95% of the correspondences are outliers.
http://arxiv.org/abs/1905.12536
In this preliminary report, we present a simple but very effective technique to stabilize the training of CNN based GANs. Motivated by recently published methods using frequency decomposition of convolutions (e.g. Octave Convolutions), we propose a novel convolution scheme to stabilize the training and reduce the likelihood of a mode collapse. The basic idea of our approach is to split convolutional filters into additive high and low frequency parts, while shifting weight updates from low to high during the training. Intuitively, this method forces GANs to learn low frequency coarse image structures before descending into fine (high frequency) details. Our approach is orthogonal and complementary to existing stabilization methods and can simply plugged into any CNN based GAN architecture. First experiments on the CelebA dataset show the effectiveness of the proposed method.
http://arxiv.org/abs/1905.12534
Adversarial examples that fool machine learning models, particularly deep neural networks, have been a topic of intense research interest, with attacks and defenses being developed in a tight back-and-forth. Most past defenses are best effort and have been shown to be vulnerable to sophisticated attacks. Recently a set of certified defenses have been introduced, which provide guarantees of robustness to norm-bounded attacks, but they either do not scale to large datasets or are limited in the types of models they can support. This paper presents the first certified defense that both scales to large networks and datasets (such as Google’s Inception network for ImageNet) and applies broadly to arbitrary model types. Our defense, called PixelDP, is based on a novel connection between robustness against adversarial examples and differential privacy, a cryptographically-inspired formalism, that provides a rigorous, generic, and flexible foundation for defense.
http://arxiv.org/abs/1802.03471
Technologies for abusive language detection are being developed and applied with little consideration of their potential biases. We examine racial bias in five different sets of Twitter data annotated for hate speech and abusive language. We train classifiers on these datasets and compare the predictions of these classifiers on tweets written in African-American English with those written in Standard American English. The results show evidence of systematic racial bias in all datasets, as classifiers trained on them tend to predict that tweets written in African-American English are abusive at substantially higher rates. If these abusive language detection systems are used in the field they will therefore have a disproportionate negative impact on African-American social media users. Consequently, these systems may discriminate against the groups who are often the targets of the abuse we are trying to detect.
http://arxiv.org/abs/1905.12516
We propose a novel 3D shape correspondence method based on the iterative alignment of so-called smooth shells. Smooth shells define a series of coarse-to-fine, smooth shape approximations that are designed to work well with multiscale algorithms. In this paper, we alternate between aligning smooth shells and computing Functional Maps between the inputs. Aligning very smooth approximations reduces the complexity of the overall process but during the iterations the amount of detail in the shells increases which helps to refine the resulting correspondence. Furthermore, we solve the problem of ambiguities from intrinsic symmetries by applying a surrogate based Markov chain Monte Carlo initialization. We show state-of-the-art quantitative results on several datasets focussing on isometries, topological changes and different connectivity. Additionally, we show qualitative results on challenging interclass pairs.
http://arxiv.org/abs/1905.12512
A disentangled representation encodes information about the salient factors of variation in the data independently. Although it is often argued that this representational format is useful in learning to solve many real-world up-stream tasks, there is little empirical evidence that supports this claim. In this paper, we conduct a large-scale study that investigates whether disentangled representations are more suitable for abstract reasoning tasks. Using two new tasks similar to Raven’s Progressive Matrices, we evaluate the usefulness of the representations learned by 360 state-of-the-art unsupervised disentanglement models. Based on these representations, we train 3600 abstract reasoning models and observe that disentangled representations do in fact lead to better up-stream performance. In particular, they appear to enable quicker learning using fewer samples.
http://arxiv.org/abs/1905.12506
In this paper, we propose GlyphGAN: style-consistent font generation based on generative adversarial networks (GANs). GANs are a framework for learning a generative model using a system of two neural networks competing with each other. One network generates synthetic images from random input vectors, and the other discriminates between synthetic and real images. The motivation of this study is to create new fonts using the GAN framework while maintaining style consistency over all characters. In GlyphGAN, the input vector for the generator network consists of two vectors: character class vector and style vector. The former is a one-hot vector and is associated with the character class of each sample image during training. The latter is a uniform random vector without supervised information. In this way, GlyphGAN can generate an infinite variety of fonts with the character and style independently controlled. Experimental results showed that fonts generated by GlyphGAN have style consistency and diversity different from the training images without losing their legibility.
http://arxiv.org/abs/1905.12502
The young field of AI Safety is still in the process of identifying its challenges and limitations. In this paper, we formally describe one such impossibility result, namely Unpredictability of AI. We prove that it is impossible to precisely and consistently predict what specific actions a smarter-than-human intelligent system will take to achieve its objectives, even if we know terminal goals of the system. In conclusion, impact of Unpredictability on AI Safety is discussed.
http://arxiv.org/abs/1905.13053
Image translation across different domains has attracted much attention in both machine learning and computer vision communities. Taking the translation from source domain $\mathcal{D}_s$ to target domain $\mathcal{D}_t$ as an example, existing algorithms mainly rely on two kinds of loss for training: One is the discrimination loss, which is used to differentiate images generated by the models and natural images; the other is the reconstruction loss, which measures the difference between an original image and the reconstructed version through $\mathcal{D}_s\to\mathcal{D}_t\to\mathcal{D}_s$ translation. In this work, we introduce a new kind of loss, multi-path consistency loss, which evaluates the differences between direct translation $\mathcal{D}_s\to\mathcal{D}_t$ and indirect translation $\mathcal{D}_s\to\mathcal{D}_a\to\mathcal{D}_t$ with $\mathcal{D}_a$ as an auxiliary domain, to regularize training. For multi-domain translation (at least, three) which focuses on building translation models between any two domains, at each training iteration, we randomly select three domains, set them respectively as the source, auxiliary and target domains, build the multi-path consistency loss and optimize the network. For two-domain translation, we need to introduce an additional auxiliary domain and construct the multi-path consistency loss. We conduct various experiments to demonstrate the effectiveness of our proposed methods, including face-to-face translation, paint-to-photo translation, and de-raining/de-noising translation.
http://arxiv.org/abs/1905.12498
In computer vision research, especially when novel applications of tools are developed, ethical implications around user perceptions of trust in the underlying technology should be considered and supported. Here, we describe an example of the incorporation of such considerations within the long-term care sector for tracking resident food and fluid intake. We highlight our recent user study conducted to develop a Goldilocks quality horizontal prototype designed to support trust cues in which perceived trust in our horizontal prototype was higher than the existing system in place. We discuss the importance and need for user engagement as part of ongoing computer vision-driven technology development and describe several important factors related to trust that are relevant to developing decision-making tools.
http://arxiv.org/abs/1905.12487
Existing review-based recommendation methods usually use the same model to learn the representations of all users/items from reviews posted by users towards items. However, different users have different preference and different items have different characteristics. Thus, the same word or similar reviews may have different informativeness for different users and items. In this paper we propose a neural recommendation approach with personalized attention to learn personalized representations of users and items from reviews. We use a review encoder to learn representations of reviews from words, and a user/item encoder to learn representations of users or items from reviews. We propose a personalized attention model, and apply it to both review and user/item encoders to select different important words and reviews for different users/items. Experiments on five datasets validate our approach can effectively improve the performance of neural recommendation.
http://arxiv.org/abs/1905.12480
In Case-Based Reasoning, when the similarity assumption does not hold, the retrieval of a set of cases structurally similar to the query does not guarantee to get a reusable or revisable solution. Knowledge about the adaptability of solutions has to be exploited, in order to define a method for adaptation-guided retrieval. We propose a novel approach to address this problem, where knowledge about the adaptability of the solutions is captured inside a metric Markov Random Field (MRF). Nodes of the MRF represent cases and edges connect nodes whose solutions are close in the solution space. States of the nodes represent different adaptation levels with respect to the potential query. Metric-based potentials enforce connected nodes to share the same state, since cases having similar solutions should have the same adaptability level with respect to the query. The main goal is to enlarge the set of potentially adaptable cases that are retrieved without significantly sacrificing the precision and accuracy of retrieval. We will report on some experiments concerning a retrieval architecture where a simple kNN retrieval (on the problem description) is followed by a further retrieval step based on MRF inference.
http://arxiv.org/abs/1905.12464
Most action recognition methods base on a) a late aggregation of frame level CNN features using average pooling, max pooling, or RNN, among others, or b) spatio-temporal aggregation via 3D convolutions. The first assume independence among frame features up to a certain level of abstraction and then perform higher-level aggregation, while the second extracts spatio-temporal features from grouped frames as early fusion. In this paper we explore the space in between these two, by letting adjacent feature branches interact as they develop into the higher level representation. The interaction happens between feature differencing and averaging at each level of the hierarchy, and it has convolutional structure that learns to select the appropriate mode locally in contrast to previous works that impose one of the modes globally (e.g. feature differencing) as a design choice. We further constrain this interaction to be conservative, e.g. a local feature subtraction in one branch is compensated by the addition on another, such that the total feature flow is preserved. We evaluate the performance of our proposal on a number of existing models, i.e. TSN, TRN and ECO, to show its flexibility and effectiveness in improving action recognition performance.
http://arxiv.org/abs/1905.12462
In this paper we present a framework that allows the motion control of a robotic arm automatically handling different kinds of safety-related tasks. The developed controller is based on a Task-Priority Inverse Kinematics algorithm that allows the manipulator’s motion while respecting constraints defined either in the joint or in the operational space in the form of equality-based or set-based tasks. This gives the possibility to define, among the others, tasks as joint-limits, obstacle avoidance or limiting the workspace in the operational space. Additionally, an algorithm for the real-time computation of the minimum distance between the manipulator and other objects in the environment using depth measurements has been implemented, effectively allowing obstacle avoidance tasks. Experiments with a Jaco$^2$ manipulator, operating in an environment where an RGB-D sensor is used for the obstacles detection, show the effectiveness of the developed system.
http://arxiv.org/abs/1905.12459
Radiomics is a rapidly growing field that deals with modeling the textural information present in the different tissues of interest for clinical decision support. However, the process of generating radiomic images is computationally very expensive and could take substantial time per radiological image for certain higher order features, such as, gray-level co-occurrence matrix(GLCM), even with high-end GPUs. To that end, we developed RadSynth, a deep convolutional neural network(CNN) model, to efficiently generate radiomic images. RadSynth was tested on a breast cancer patient cohort of twenty-four patients(ten benign, ten malignant and four normal) for computation of GLCM entropy images from post-contrast DCE-MRI. RadSynth produced excellent synthetic entropy images compared to traditional GLCM entropy images. The average percentage difference and correlation between the two techniques were 0.07 $\pm$ 0.06 and 0.97, respectively. In conclusion, RadSynth presents a new powerful tool for fast computation and visualization of the textural information present in the radiological images.
http://arxiv.org/abs/1810.11090
In recent years, precision agriculture that uses modern information and communication technologies is becoming very popular. Raw and semi-processed agricultural data are usually collected through various sources, such as: Internet of Thing (IoT), sensors, satellites, weather stations, robots, farm equipment, farmers and agribusinesses, etc. Besides, agricultural datasets are very large, complex, unstructured, heterogeneous, non-standardized, and inconsistent. Hence, the agricultural data mining is considered as Big Data application in terms of volume, variety, velocity and veracity. It is a key foundation to establishing a crop intelligence platform, which will enable resource efficient agronomy decision making and recommendations. In this paper, we designed and implemented a continental level agricultural data warehouse by combining Hive, MongoDB and Cassandra. Our data warehouse capabilities: (1) flexible schema; (2) data integration from real agricultural multi datasets; (3) data science and business intelligent support; (4) high performance; (5) high storage; (6) security; (7) governance and monitoring; (8) replication and recovery; (9) consistency, availability and partition tolerant; (10) distributed and cloud deployment. We also evaluate the performance of our data warehouse.
http://arxiv.org/abs/1905.12411
In tracking of time-varying low-rank models of time-varying matrices, we present a method robust to both uniformly-distributed measurement noise and arbitrarily-distributed “sparse” noise. In theory, we bound the tracking error. In practice, our use of randomised coordinate descent is scalable and allows for encouraging results on a benchmark (changedetection net).
http://arxiv.org/abs/1809.03550
Multi-Person Tracking (MPT) is often addressed within the detection-to-association paradigm. In such approaches, human detections are first extracted in every frame and person trajectories are then recovered by a procedure of data association (usually offline). However, their performances usually degenerate in presence of detection errors, mutual interactions and occlusions. In this paper, we present a deep learning based MPT approach that learns instance-aware representations of tracked persons and robustly online infers states of the tracked persons. Specifically, we design a multi-branch neural network (MBN), which predicts the classification confidences and locations of all targets by taking a batch of candidate regions as input. In our MBN architecture, each branch (instance-subnet) corresponds to an individual to be tracked and new branches can be dynamically created for handling newly appearing persons. Then based on the output of MBN, we construct a joint association matrix that represents meaningful states of tracked persons (e.g., being tracked or disappearing from the scene) and solve it by using the efficient Hungarian algorithm. Moreover, we allow the instance-subnets to be updated during tracking by online mining hard examples, accounting to person appearance variations over time. We comprehensively evaluate our framework on a popular MPT benchmark, demonstrating its excellent performance in comparison with recent online MPT methods.
http://arxiv.org/abs/1905.12409
We propose a set of compositional design patterns to describe a large variety of systems that combine statistical techniques from machine learning with symbolic techniques from knowledge representation. As in other areas of computer science (knowledge engineering, software engineering, ontology engineering, process mining and others), such design patterns help to systematize the literature, clarify which combinations of techniques serve which purposes, and encourage re-use of software components. We have validated our set of compositional design patterns against a large body of recent literature.
http://arxiv.org/abs/1905.12389
The latest deep learning-based approaches have shown promising results for the challenging task of inpainting missing regions of an image. However, the existing methods often generate contents with blurry textures and distorted structures due to the discontinuity of the local pixels. From a semantic-level perspective, the local pixel discontinuity is mainly because these methods ignore the semantic relevance and feature continuity of hole regions. To handle this problem, we investigate the human behavior in repairing pictures and propose a fined deep generative model-based approach with a novel coherent semantic attention (CSA) layer, which can not only preserve contextual structure but also make more effective predictions of missing parts by modeling the semantic relevance between the holes features. The task is divided into rough, refinement as two steps and model each step with a neural network under the U-Net architecture, where the CSA layer is embedded into the encoder of refinement step. To stabilize the network training process and promote the CSA layer to learn more effective parameters, we propose a consistency loss to enforce the both the CSA layer and the corresponding layer of the CSA in decoder to be close to the VGG feature layer of a ground truth image simultaneously. The experiments on CelebA, Places2, and Paris StreetView datasets have validated the effectiveness of our proposed methods in image inpainting tasks and can obtain images with a higher quality as compared with the existing state-of-the-art approaches.
http://arxiv.org/abs/1905.12384
In this paper we propose an approach for monocular 3D object detection from a single RGB image, which leverages a novel disentangling transformation for 2D and 3D detection losses and a novel, self-supervised confidence score for 3D bounding boxes. Our proposed loss disentanglement has the twofold advantage of simplifying the training dynamics in the presence of losses with complex interactions of parameters, and sidestepping the issue of balancing independent regression terms. Our solution overcomes these issues by isolating the contribution made by groups of parameters to a given loss, without changing its nature. We further apply loss disentanglement to another novel, signed Intersection-over-Union criterion-driven loss for improving 2D detection results. Besides our methodological innovations, we critically review the AP metric used in KITTI3D, which emerged as the most important dataset for comparing 3D detection results. We identify and resolve a flaw in the 11-point interpolated AP metric, affecting all previously published detection results and particularly biases the results of monocular 3D detection. We provide extensive experimental evaluations and ablation studies on the KITTI3D and nuScenes datasets, setting new state-of-the-art results on object category car by large margins.
http://arxiv.org/abs/1905.12365
Supervised machine learning (ML) algorithms are aimed at maximizing classification performance under available energy and storage constraints. They try to map the training data to the corresponding labels while ensuring generalizability to unseen data. However, they do not integrate meaning-based relationships among labels in the decision process. On the other hand, natural language processing (NLP) algorithms emphasize the importance of semantic information. In this paper, we synthesize the complementary advantages of supervised ML and natural language processing algorithms into one method that we refer to as SECRET (Semantically Enhanced Classification of REal-world Tasks). SECRET performs classifications by fusing the semantic information of the labels with the available data: it combines the feature space of the supervised algorithms with the semantic space of the NLP algorithms and predicts labels based on this joint space. Experimental results indicate that, compared to traditional supervised learning, SECRET achieves up to 13.9% accuracy and 13.5% F1 score improvements. Moreover, compared to ensemble methods, SECRET achieves up to 12.6% accuracy and 13.8% F1 score improvements. This points to a new research direction for supervised classification by incorporating semantic information.
http://arxiv.org/abs/1905.12356
Recent studies have demonstrated that the convolutional networks heavily rely on the quality and quantity of generated features. However, in lightweight networks, there are limited available feature information because these networks tend to be shallower and thinner due to the efficiency consideration. For farther improving the performance and accuracy of lightweight networks, we develop Super Interaction Neural Networks (SINet) model from a novel point of view: enhancing the information interaction in neural networks. In order to achieve information interaction along the width of the deep network, we propose Exchange Shortcut Connection, which can integrate the information from different convolution groups without any extra computation cost. And then, in order to achieve information interaction along the depth of the network, we proposed Dense Funnel Layer and Attention based Hierarchical Joint Decision, which are able to make full use of middle layer features. Our experiments show that the superior performance of SINet over other state-of-the-art lightweight models in ImageNet dataset. Furthermore, we also exhibit the effectiveness and universality of our proposed components by ablation studies.
http://arxiv.org/abs/1905.12349