We consider the problem of designing an artificial agent capable of interacting with humans in collaborative dialogue to produce creative, engaging narratives. In this task, the goal is to establish universe details, and to collaborate on an interesting story in that universe, through a series of natural dialogue exchanges. Our model can augment any probabilistic conversational agent by allowing it to reason about universe information established and what potential next utterances might reveal. Ideally, with each utterance, agents would reveal just enough information to add specificity and reduce ambiguity without limiting the conversation. We empirically show that our model allows control over the rate at which the agent reveals information and that doing so significantly improves accuracy in predicting the next line of dialogues from movies. We close with a case-study with four professional theatre performers, who preferred interactions with our model-augmented agent over an unaugmented agent.
http://arxiv.org/abs/1901.11528
We establish geometric and topological properties of the space of value functions in finite state-action Markov decision processes. Our main contribution is the characterization of the nature of its shape: a general polytope (Aigner et al., 2010). To demonstrate this result, we exhibit several properties of the structural relationship between policies and value functions including the line theorem, which shows that the value functions of policies constrained on all but one state describe a line segment. Finally, we use this novel perspective to introduce visualizations to enhance the understanding of the dynamics of reinforcement learning algorithms.
http://arxiv.org/abs/1901.11524
Optimizing an expensive-to-query function is a common task in science and engineering, where it is beneficial to keep the number of queries to a minimum. A popular strategy is Bayesian optimization (BO), which leverages probabilistic models for this task. Most BO today uses Gaussian processes (GPs), or a few other surrogate models. However, there is a broad set of Bayesian modeling techniques that we may want to use to capture complex systems and reduce the number of queries. Probabilistic programs (PPs) are modern tools that allow for flexible model composition, incorporation of prior information, and automatic inference. In this paper, we develop ProBO, a framework for BO using only standard operations common to most PPs. This allows a user to drop in an arbitrary PP implementation and use it directly in BO. To do this, we describe black box versions of popular acquisition functions that can be used in our framework automatically, without model-specific derivation, and show how to optimize these functions. We also introduce a model, which we term the Bayesian Product of Experts, that integrates into ProBO and can be used to combine information from multiple models implemented with different PPs. We show empirical results using multiple PP implementations, and compare against standard BO methods.
http://arxiv.org/abs/1901.11515
In this paper, we present a Multi-Task Deep Neural Network (MT-DNN) for learning representations across multiple natural language understanding (NLU) tasks. MT-DNN not only leverages large amounts of cross-task data, but also benefits from a regularization effect that leads to more general representations in order to adapt to new tasks and domains. MT-DNN extends the model proposed in Liu et al. (2015) by incorporating a pre-trained bidirectional transformer language model, known as BERT (Devlin et al., 2018). MT-DNN obtains new state-of-the-art results on ten NLU tasks, including SNLI, SciTail, and eight out of nine GLUE tasks, pushing the GLUE benchmark to 82.2% (1.8% absolute improvement). We also demonstrate using the SNLI and SciTail datasets that the representations learned by MT-DNN allow domain adaptation with substantially fewer in-domain labels than the pre-trained BERT representations. Our code and pre-trained models will be made publicly available.
http://arxiv.org/abs/1901.11504
Black-box optimizers that explore in parameter space have often been shown to outperform more sophisticated action space exploration methods developed specifically for the reinforcement learning problem. We examine these black-box methods closely to identify situations in which they are worse than action space exploration methods and those in which they are superior. Through simple theoretical analyses, we prove that complexity of exploration in parameter space depends on the dimensionality of parameter space, while complexity of exploration in action space depends on both the dimensionality of action space and horizon length. This is also demonstrated empirically by comparing simple exploration methods on several model problems, including Contextual Bandit, Linear Regression and Reinforcement Learning in continuous control.
http://arxiv.org/abs/1901.11503
Enabling autonomous robots to interact in unstructured environments with dynamic objects requires manipulation capabilities that can deal with clutter, changes, and objects’ variability. This paper presents a comparison of different reinforcement learning-based approaches for object picking with a robotic manipulator. We learn closed-loop policies mapping depth camera inputs to motion commands and compare different approaches to keep the problem tractable, including reward shaping, curriculum learning and using a policy pre-trained on a task with a reduced action set to warm-start the full problem. For efficient and more flexible data collection, we train in simulation and transfer the policies to a real robot. We show that using curriculum learning, policies learned with a sparse reward formulation can be trained at similar rates as with a shaped reward. These policies result in success rates comparable to the policy initialized on the simplified task. We could successfully transfer these policies to the real robot with only minor modifications of the depth image filtering. We found that using a heuristic to warm-start the training was useful to enforce desired behavior, while the policies trained from scratch using a curriculum learned better to cope with unseen scenarios where objects are removed.
http://arxiv.org/abs/1803.04996
Classification of histologic patterns in lung adenocarcinoma is critical for determining tumor grade and treatment for patients. However, this task is often challenging due to the heterogeneous nature of lung adenocarcinoma and the subjective criteria for evaluation. In this study, we propose a deep learning model that automatically classifies the histologic patterns of lung adenocarcinoma on surgical resection slides. Our model uses a convolutional neural network to identify regions of neoplastic cells, then aggregates those classifications to infer predominant and minor histologic patterns for any given whole-slide image. We evaluated our model on an independent set of 143 whole-slide images. It achieved a kappa score of 0.525 and an agreement of 66.6% with three pathologists for classifying the predominant patterns, slightly higher than the inter-pathologist kappa score of 0.485 and agreement of 62.7% on this test set. All evaluation metrics for our model and the three pathologists were within 95% confidence intervals of agreement. If confirmed in clinical practice, our model can assist pathologists in improving classification of lung adenocarcinoma patterns by automatically pre-screening and highlighting cancerous regions prior to review. Our approach can be generalized to any whole-slide image classification task, and code is made publicly available at https://github.com/BMIRDS/deepslide.
http://arxiv.org/abs/1901.11489
The era of Unmanned aviation is flourishing and advancing in leaps and bounds jumping over all challenges and obstacles. One of the serious hinderances to the booming industry is the lack of sufficient reliability studies about the systems used on board these flying robots. Accordingly, this poses a threat to the safety and well-being of human communities as well as the loss of assets, especially in BLOS flights. The need for a reliable endurance investigation is emerging from the fact that many of the systems and components used on UAS are COTS which comes with scarce information, if any, about its reliability and operation-life expectancy. This topic is of vital importance to regulatory organizations to consider legalizing commercial flights that are not in the field of the human operators’ line of sight. Servos are one type of these crucial components used in UAS which lack thorough reliability evaluations. This study addresses this concern through a case study of the servo used on the Boeing Insitu ScanEagle to control its ailerons. To offer the expected information about the reliability of the components, a destructive test platform was sought. A test-bed was designed to operate the servo to failure using the actual commanded positions it operates to in its real-life service. The test-bed commands the servo and logs its actual movements along with the commanded positions. These logs are used to detect the discrepancy between the commanded and actual positions to give a statistical estimate about how long the servo will endure.
http://arxiv.org/abs/1901.11486
We consider discrete dynamical systems of “ant-like” agents engaged in a sequence of pursuits on a graph environment. The agents emerge one by one at equal time intervals from a source vertex $s$ and pursue each other by greedily attempting to close the distance to their immediate predecessor, the agent that emerged just before them from $s$, until they arrive at the destination point $t$. Such pursuits have been investigated before in the continuous setting and in discrete time when the underlying environment is a regular grid. In both these settings the agents’ walks provably converge to a shortest path from $s$ to $t$. Furthermore, assuming a certain natural probability distribution over the move choices of the agents on the grid (in case there are multiple shortest paths between an agent and its predecessor), the walks converge to the uniform distribution over all shortest paths from $s$ to $t$. We study the evolution of agent walks over a general finite graph environment $G$. Our model is a natural generalization of the pursuit rule proposed for the case of the grid. The main results are as follows. We show that “convergence” to the shortest paths in the sense of previous work extends to all pseudo-modular graphs (i.e. graphs in which every three pairwise intersecting disks have a nonempty intersection), and also to environments obtained by taking graph products, generalizing previous results in two different ways. We show that convergence to the shortest paths is also obtained by chordal graphs, and discuss some further positive and negative results for planar graphs. In the most general case, convergence to the shortest paths is not guaranteed, and the agents may get stuck on sets of recurrent, non-optimal walks from $s$ to $t$. However, we show that the limiting distributions of the agents’ walks will always be uniform distributions over some set of walks of equal length.
http://arxiv.org/abs/1710.08107
Taxonomies are semantic hierarchies of concepts. One limitation of current taxonomy learning systems is that they define concepts as single words. This position paper argues that contextualized word representations, which recently achieved state-of-the-art results on many competitive NLP tasks, are a promising method to address this limitation. We outline a novel approach for taxonomy learning that (1) defines concepts as synsets, (2) learns density-based approximations of contextualized word representations, and (3) can measure similarity and hypernymy among them.
http://arxiv.org/abs/1902.02169
Curriculum learning is gaining popularity in (deep) reinforcement learning. It can alleviate the burden on data collection and provide better exploration policies through transfer and generalization from less complex tasks. Current methods for automatic task sequencing for curriculum learning in reinforcement learning provided initial heuristic solutions, with little to no guarantee on their quality. We introduce an optimization framework for task sequencing composed of the problem definition, several candidate performance metrics for optimization, and three benchmark algorithms. We experimentally show that the two most commonly used baselines (learning with no curriculum, and with a random curriculum) perform worse than a simple greedy algorithm. Furthermore, we show theoretically and demonstrate experimentally that the three proposed algorithms provide increasing solution quality at moderately increasing computational complexity, and show that they constitute better baselines for curriculum learning in reinforcement learning.
http://arxiv.org/abs/1901.11478
An obstacle to the development of many natural language processing products is the vast amount of training examples necessary to get satisfactory results. The generation of these examples is often a tedious and time-consuming task. This paper this paper proposes a method to transform the sentiment of sentences in order to limit the work necessary to generate more training data. This means that one sentence can be transformed to an opposite sentiment sentence and should reduce by half the work required in the generation of text. The proposed pipeline consists of a sentiment classifier with an attention mechanism to highlight the short phrases that determine the sentiment of a sentence. Then, these phrases are changed to phrases of the opposite sentiment using a baseline model and an autoencoder approach. Experiments are run on both the separate parts of the pipeline as well as on the end-to-end model. The sentiment classifier is tested on its accuracy and is found to perform adequately. The autoencoder is tested on how well it is able to change the sentiment of an encoded phrase and it was found that such a task is possible. We use human evaluation to judge the performance of the full (end-to-end) pipeline and that reveals that a model using word vectors outperforms the encoder model. Numerical evaluation shows that a success rate of 54.7% is achieved on the sentiment change.
http://arxiv.org/abs/1901.11467
Conversational agents have begun to rise both in the academic (in terms of research) and commercial (in terms of applications) world. This paper investigates the task of building a non-goal driven conversational agent, using neural network generative models and analyzes how the conversation context is handled. It compares a simpler Encoder-Decoder with a Hierarchical Recurrent Encoder-Decoder architecture, which includes an additional module to model the context of the conversation using previous utterances information. We found that the hierarchical model was able to extract relevant context information and include them in the generation of the output. However, it performed worse (35-40%) than the simple Encoder-Decoder model regarding both grammatically correct output and meaningful response. Despite these results, experiments demonstrate how conversations about similar topics appear close to each other in the context space due to the increased frequency of specific topic-related words, thus leaving promising directions for future research and how the context of a conversation can be exploited.
http://arxiv.org/abs/1901.11462
Mesh models are a promising approach for encoding the structure of 3D objects. Current mesh reconstruction systems predict uniformly distributed vertex locations of a predetermined graph through a series of graph convolutions, leading to compromises with respect to performance or resolution. In this paper, we argue that the graph representation of geometric objects allows for additional structure, which should be leveraged for enhanced reconstruction. Thus, we propose a system which properly benefits from the advantages of the geometric structure of graph encoded objects by introducing (1) a graph convolutional update preserving vertex information; (2) an adaptive splitting heuristic allowing detail to emerge; and (3) a training objective operating both on the local surfaces defined by vertices as well as the global structure defined by the mesh. Our proposed method is evaluated on the task of 3D object reconstruction from images with the ShapeNet dataset, where we demonstrate state of the art performance, both visually and numerically, while having far smaller space requirements by generating adaptive meshes
http://arxiv.org/abs/1901.11461
Polylingual Text Classification (PLC) consists of automatically classifying, according to a common set C of classes, documents each written in one of a set of languages L, and doing so more accurately than when naively classifying each document via its corresponding language-specific classifier. In order to obtain an increase in the classification accuracy for a given language, the system thus needs to also leverage the training examples written in the other languages. We tackle multilabel PLC via funnelling, a new ensemble learning method that we propose here. Funnelling consists of generating a two-tier classification system where all documents, irrespectively of language, are classified by the same (2nd-tier) classifier. For this classifier all documents are represented in a common, language-independent feature space consisting of the posterior probabilities generated by 1st-tier, language-dependent classifiers. This allows the classification of all test documents, of any language, to benefit from the information present in all training documents, of any language. We present substantial experiments, run on publicly available polylingual text collections, in which funnelling is shown to significantly outperform a number of state-of-the-art baselines. All code and datasets (in vector form) are made publicly available.
http://arxiv.org/abs/1901.11459
The work presents two algorithms of manipulation and comparison between strings whose purpose is the orthographic recognition of the apostrophe and of the compound expressions. The theory supporting general reasoning refers to the basic concept of EditDistance, the improvements that ensure the achievement of the objective are achieved with the aid of tools borrowed from the use of techniques for processing large amounts of data on distributed platforms.
http://arxiv.org/abs/1902.00555
Celiac disease prevalence and diagnosis have increased substantially in recent years. The current gold standard for celiac disease confirmation is visual examination of duodenal mucosal biopsies. An accurate computer-aided biopsy analysis system using deep learning can help pathologists diagnose celiac disease more efficiently. In this study, we trained a deep learning model to detect celiac disease on duodenal biopsy images. Our model uses a state-of-the-art residual convolutional neural network to evaluate patches of duodenal tissue and then aggregates those predictions for whole-slide classification. We tested the model on an independent set of 212 images and evaluated its classification results against reference standards established by pathologists. Our model identified celiac disease, normal tissue, and nonspecific duodenitis with accuracies of 95.3%, 91.0%, and 89.2%, respectively. The area under the receiver operating characteristic curve was greater than 0.95 for all classes. We have developed an automated biopsy analysis system that achieves high performance in detecting celiac disease on biopsy slides. Our system can highlight areas of interest and provide preliminary classification of duodenal biopsies prior to review by pathologists. This technology has great potential for improving the accuracy and efficiency of celiac disease diagnosis.
http://arxiv.org/abs/1901.11447
One key challenge in reinforcement learning is the ability to generalize knowledge in control problems. While deep learning methods have been successfully combined with model-free reinforcement-learning algorithms, how to perform model-based reinforcement learning in the presence of approximation errors still remains an open problem. Using successor features, a feature representation that predicts a temporal constraint, this paper presents three contributions: First, it shows how learning successor features is equivalent to model-free learning. Then, it shows how successor features encode model reductions that compress the state space by creating state partitions of bisimilar states. Using this representation, an intelligent agent is guaranteed to accurately predict future reward outcomes, a key property of model-based reinforcement-learning algorithms. Lastly, it presents a loss objective and prediction error bounds showing that accurately predicting value functions and reward sequences is possible with an approximation of successor features. On finite control problems, we illustrate how minimizing this loss objective results in approximate bisimulations. The results presented in this paper provide a novel understanding of representations that can support model-free and model-based reinforcement learning.
http://arxiv.org/abs/1901.11437
A typical audio signal processing pipeline includes multiple disjoint analysis stages, including calculation of a time-frequency representation followed by spectrogram-based feature analysis. We show how time-frequency analysis and nonnegative matrix factorisation can be jointly formulated as a spectral mixture Gaussian process model with nonstationary priors over the amplitude variance parameters. Further, we formulate this nonlinear model’s state space representation, making it amenable to infinite-horizon Gaussian process regression with approximate inference via expectation propagation, which scales linearly in the number of time steps and quadratically in the state dimensionality. By doing so, we are able to process audio signals with hundreds of thousands of data points. We demonstrate, on various tasks with empirical data, how this inference scheme outperforms more standard techniques that rely on extended Kalman filtering.
http://arxiv.org/abs/1901.11436
We present a novel semantic framework for modeling linguistic expressions of generalization - generic, habitual, and episodic statements - as combinations of simple, real-valued referential properties of predicates and their arguments. We use this framework to construct a dataset covering the entirety of the Universal Dependencies English Web Treebank. We use this dataset to probe the efficacy of type-level and token-level information - including hand-engineered features and contextual and non-contextual word embeddings - for predicting expressions of generalization. Data and code are available at decomp.io.
http://arxiv.org/abs/1901.11429
This paper deals with the prediction of the memorability of a given image. We start by proposing an algorithm that reaches human-level performance on the LaMem dataset - the only large scale benchmark for memorability prediction. The suggested algorithm is based on three observations we make regarding convolutional neural networks (CNNs) that affect memorability prediction. Having reached human-level performance we were humbled, and asked ourselves whether indeed we have resolved memorability prediction - and answered this question in the negative. We studied a few factors and made some recommendations that should be taken into account when designing the next benchmark.
http://arxiv.org/abs/1901.11420
To solve combinatorial problems, traditional search-based optimization uses heuristics to guide the search. Deciding in which specific conditions and situations a certain heuristic can be applied is time-consuming and might cost decades to tune. In this paper, we learn a policy to pick heuristics and rewrite the local components of the current solution to iteratively improve it until convergence. The policy factorizes into a region-picking policy and a rule-picking policy, and it employs a neural network trained with reinforcement learning. We evaluate our approach in three domains: expression simplification, online job scheduling and SAT. Our approach outperforms Z3 (De Moura & Bj{\o}rner, 2008), a state-of-the-art theorem prover, in expression simplification, outperforms DeepRM (Mao et al., 2016) and Google OR-tools (Google, 2019) in online job scheduling, and outperforms NeuroSAT (Selsam et al., 2019) and DG-DAGRNN (Amizadeh et al., 2019) in SAT with a small number of variables.
http://arxiv.org/abs/1810.00337
In this paper, we introduce a novel concept for learning of the parameters in a neural network. Our idea is grounded on modeling a learning problem that addresses a trade-off between (i) satisfying local objectives at each node and (ii) achieving desired data propagation through the network under (iii) local propagation constraints. We consider two types of nonlinear transforms which describe the network representations. One of the nonlinear transforms serves as activation function. The other one enables a locally adjusted, deviation corrective components to be included in the update of the network weights in order to enable attaining target specific representations at the last network node. Our learning principle not only provides insight into the understanding and the interpretation of the learning dynamics, but it offers theoretical guarantees over decoupled and parallel parameter estimation strategy that enables learning in synchronous and asynchronous mode. Numerical experiments validate the potential of our approach on image recognition task. The preliminary results show advantages in comparison to the state-of-the-art methods, w.r.t. the learning time and the network size while having competitive recognition accuracy.
http://arxiv.org/abs/1902.00016
We define general linguistic intelligence as the ability to reuse previously acquired knowledge about a language’s lexicon, syntax, semantics, and pragmatic conventions to adapt to new tasks quickly. Using this definition, we analyze state-of-the-art natural language understanding models and conduct an extensive empirical investigation to evaluate them against these criteria through a series of experiments that assess the task-independence of the knowledge being acquired by the learning process. In addition to task performance, we propose a new evaluation metric based on an online encoding of the test data that quantifies how quickly an existing agent (model) learns a new task. Our results show that while the field has made impressive progress in terms of model architectures that generalize to many tasks, these models still require a lot of in-domain training examples (e.g., for fine tuning, training task-specific modules), and are prone to catastrophic forgetting. Moreover, we find that far from solving general tasks (e.g., document question answering), our models are overfitting to the quirks of particular datasets (e.g., SQuAD). We discuss missing components and conjecture on how to make progress toward general linguistic intelligence.
http://arxiv.org/abs/1901.11373
Lack of large expert annotated MR datasets makes training deep learning models difficult. Therefore, a cross-modality (MR-CT) deep learning segmentation approach that augments training data using pseudo MR images produced by transforming expert-segmented CT images was developed. Eighty-One T2-weighted MRI scans from 28 patients with non-small cell lung cancers were analyzed. Cross-modality prior encoding the transformation of CT to pseudo MR images resembling T2w MRI was learned as a generative adversarial deep learning model. This model augmented training data arising from 6 expert-segmented T2w MR patient scans with 377 pseudo MRI from non-small cell lung cancer CT patient scans with obtained from the Cancer Imaging Archive. A two-dimensional Unet implemented with batch normalization was trained to segment the tumors from T2w MRI. This method was benchmarked against (a) standard data augmentation and two state-of-the art cross-modality pseudo MR-based augmentation and (b) two segmentation networks. Segmentation accuracy was computed using Dice similarity coefficient (DSC), Hausdroff distance metrics, and volume ratio. The proposed approach produced the lowest statistical variability in the intensity distribution between pseudo and T2w MR images measured as Kullback-Leibler divergence of 0.069. This method produced the highest segmentation accuracy with a DSC of 0.75 and the lowest Hausdroff distance on the test dataset. This approach produced highly similar estimations of tumor growth as an expert (P = 0.37). A novel deep learning MR segmentation was developed that overcomes the limitation of learning robust models from small datasets by leveraging learned cross-modality priors to augment training. The results show the feasibility of the approach and the corresponding improvement over the state-of-the-art methods.
http://arxiv.org/abs/1901.11369
Multi-layer models with multiple attention heads per layer provide superior translation quality compared to simpler and shallower models, but determining what source context is most relevant to each target word is more challenging as a result. Therefore, deriving high-accuracy word alignments from the activations of a state-of-the-art neural machine translation model is an open challenge. We propose a simple model extension to the Transformer architecture that makes use of its hidden representations and is restricted to attend solely on encoder information to predict the next word. It can be trained on bilingual data without word-alignment information. We further introduce a novel alignment inference procedure which applies stochastic gradient descent to directly optimize the attention activations towards a given target word. The resulting alignments dramatically outperform the naive approach to interpreting Transformer attention activations, and are comparable to Giza++ on two publicly available data sets.
http://arxiv.org/abs/1901.11359
We investigate the problem whether two ALC ontologies are indistinguishable (or inseparable) by means of queries in a given signature, which is fundamental for ontology engineering tasks such as ontology versioning, modularisation, update, and forgetting. We consider both knowledge base (KB) and TBox inseparability. For KBs, we give model-theoretic criteria in terms of (finite partial) homomorphisms and products and prove that this problem is undecidable for conjunctive queries (CQs), but 2ExpTime-complete for unions of CQs (UCQs). The same results hold if (U)CQs are replaced by rooted (U)CQs, where every variable is connected to an answer variable. We also show that inseparability by CQs is still undecidable if one KB is given in the lightweight DL EL and if no restrictions are imposed on the signature of the CQs. We also consider the problem whether two ALC TBoxes give the same answers to any query over any ABox in a given signature and show that, for CQs, this problem is undecidable, too. We then develop model-theoretic criteria for Horn-ALC TBoxes and show using tree automata that, in contrast, inseparability becomes decidable and 2ExpTime-complete, even ExpTime-complete when restricted to (unions of) rooted CQs.
http://arxiv.org/abs/1902.00014
We propose two minimal solutions to the problem of relative pose estimation of (i) a calibrated camera from four points in two views and (ii) a calibrated generalized camera from five points in two views. In both cases, the relative rotation angle between the views is assumed to be known. In practice, such angle can be derived from the readings of a 3d gyroscope. We represent the rotation part of the motion in terms of unit quaternions in order to construct polynomial equations encoding the epipolar constraints. The Gr"{o}bner basis technique is then used to efficiently derive the solutions. Our first solver for regular cameras significantly improves the existing state-of-the-art solution. The second solver for generalized cameras is novel. The presented minimal solvers can be used in a hypothesize-and-test architecture such as RANSAC for reliable pose estimation. Experiments on synthetic and real datasets confirm that our algorithms are numerically stable, fast and robust.
http://arxiv.org/abs/1901.11357
Recent years has witnessed dramatic progress of neural machine translation (NMT), however, the method of manually guiding the translation procedure remains to be better explored. Previous works proposed to handle such problem through lexcially-constrained beam search in the decoding phase. Unfortunately, these lexically-constrained beam search methods suffer two fatal disadvantages: high computational complexity and hard beam search which generates unexpected translations. In this paper, we propose to learn the ability of lexically-constrained translation with external memory, which can overcome the above mentioned disadvantages. For the training process, automatically extracted phrase pairs are extracted from alignment and sentence parsing, then further be encoded into an external memory. This memory is then used to provide lexically-constrained information for training through a memory-attention machanism. Various experiments are conducted on WMT Chinese to English and English to German tasks. All the results can demonstrate the effectiveness of our method.
http://arxiv.org/abs/1901.11344
Brain extraction is a critical preprocessing step in the analysis of MRI neuroimaging studies and influences the accuracy of downstream analyses. State-of-the-art brain extraction algorithms are, however, optimized for processing healthy brains and thus frequently fail in the presence of pathologically altered brain or when applied to heterogeneous MRI datasets. Here we introduce a new, rigorously validated algorithm (termed HD-BET) relying on artificial neural networks that aims to overcome these limitations. We demonstrate that HD-BET outperforms five publicly available state-of-the-art brain extraction algorithms in several large-scale neuroimaging datasets, including one from a prospective multicentric trial in neuro-oncology, yielding median improvements of +1.33 to +2.63 points for the DICE coefficient and -0.80 to -2.75 mm for the Hausdorff distance (Bonferroni-adjusted p<0.001). Importantly, the HD-BET algorithm shows robust performance in the presence of pathology or treatment-induced tissue alterations, is applicable to a broad range of MRI sequence types and is not influenced by variations in MRI hardware and acquisition parameters encountered in both research and clinical practice. For broader accessibility our HD-BET prediction algorithm is made freely available and may become an essential component for robust, automated, high-throughput processing of MRI neuroimaging data.
http://arxiv.org/abs/1901.11341
Text style transfer seeks to learn how to automatically rewrite sentences from a source domain to the target domain in different styles, while simultaneously preserving their semantic contents. A major challenge in this task stems from the lack of parallel data that connects the source and target styles. Existing approaches try to disentangle content and style, but this is quite difficult and often results in poor content-preservation and grammaticality. In contrast, we propose a novel approach by first constructing a pseudo-parallel resource that aligns a subset of sentences with similar content between source and target corpus. And then a standard sequence-to-sequence model can be applied to learn the style transfer. Subsequently, we iteratively refine the learned style transfer function while improving upon the imperfections in our original alignment. Our method is applied to the tasks of sentiment modification and formality transfer, where it outperforms state-of-the-art systems by a large margin. As an auxiliary contribution, we produced a publicly-available test set with human-generated style transfers for future community use.
http://arxiv.org/abs/1901.11333
This paper explores two techniques to improve the performance of text-dependent speaker verification systems based on deep neural networks. Firstly, we propose a general alignment mechanism to keep the temporal structure of each phrase and obtain a supervector with the speaker and phrase information, since both are relevant for a text-dependent verification. As we show, it is possible to use different alignment techniques to replace the average pooling providing significant gains in performance. Moreover, we present a novel back-end approach to train a neural network for detection tasks by optimizing the Area Under the Curve (AUC) as an alternative to the usual triplet loss function, so the system is end-to-end, with a cost function closed to our desired measure of performance. As we can see in the experimental section, this approach improves the system performance, since our triplet AUC neural network learns how to discriminate between pairs of examples from the same identity and pairs of different identities. The different alignment techniques to produce supervectors in addition to the new back-end approach were tested on the RSR2015-Part I database for text-dependent speaker verification, providing competitive results compared to similar size networks using the average pooling to extract supervectors and using a simple back-end or triplet loss training.
http://arxiv.org/abs/1901.11332
We look into robustness of deep learning based MRI reconstruction when tested on unseen contrasts and organs. We then propose to generalise the network by training with large publicly-available natural image datasets with synthesised phase information to achieve high cross-domain reconstruction performance which is competitive with domain-specific training. To explain its generalisation mechanism, we have also analysed patch sets for different training datasets.
http://arxiv.org/abs/1902.10815
In this work, we address a novel, but potentially emerging, problem of discriminating the natural human voices and those played back by any kind of audio devices in the context of interactions with in-house voice user interface. The tackled problem may find relevant applications in (1) the far-field voice interactions of vocal interfaces such as Amazon Echo, Google Home, Facebook Portal, etc, and (2) the replay spoofing attack detection. The detection of loudspeaker emitted speech will help avoid false wake-ups or unintended interactions with the devices in the first application, while eliminating attacks involve the replay of recordings collected from enrolled speakers in the second one. At first we collect a real-world dataset under well-controlled conditions containing two classes: recorded speeches directly spoken by numerous people (considered as the natural speech), and recorded speeches played back from various loudspeakers (considered as the loudspeaker emitted speech). Then from this dataset, we build prediction models based on Deep Neural Network (DNN) for which different combination of audio features have been considered. Experiment results confirm the feasibility of the task where the combination of audio embeddings extracted from SoundNet and VGGish network yields the classification accuracy up to about 90%.
http://arxiv.org/abs/1901.11291
We propose a deep convolutional object detector for automated driving applications that also estimates classification, pose and shape uncertainty of each detected object. The input consists of a multi-layer grid map which is well-suited for sensor fusion, free-space estimation and machine learning. Based on the estimated pose and shape uncertainty we approximate object hulls with bounded collision probability which we find helpful for subsequent trajectory planning tasks. We train our models based on the KITTI object detection data set. In a quantitative and qualitative evaluation some models show a similar performance and superior robustness compared to previously developed object detectors. However, our evaluation also points to undesired data set properties which should be addressed when training data-driven models or creating new data sets.
http://arxiv.org/abs/1901.11284
The Multi-agent Path Finding (MAPF) problem consists of all agents having to move to their own destinations while avoiding collisions. In practical applications of the problem, such as for navigation in an automated warehouse, MAPF must be solved iteratively. We present a novel approach for iterative MAPF, that we call Priority Inheritance with Backtracking (PIBT). In our method, a unique priority is assigned to each agent every timestep, so that movements are prioritized. Priority inheritance aims to deal effectively with priority inversion in path adjustment in a small time window. When a low-priority agent X impedes the movement of a higher-priority agent Y, agent X inherits the higher-priority of agent Y. Priority inheritance can be applied iteratively and the backtracking protocol prevents the presence of a stuck agent. We proved that, regardless of the number of agents, all agents are guaranteed to reach their own destinations within a finite time after the destinations are given, as long as the environment is a biconnected graph. Our implementation of PIBT can be fully decentralized in the sense that it has no single point of control and does not rely on global communication (all communication is local within 2-hops). Experimental results over various domains demonstrate that the practicality of the proposed algorithm which is shown to produce sophisticated coordination.
http://arxiv.org/abs/1901.11282
Moderation of user-generated content in an online community is a challenge that has great socio-economical ramifications. However, the costs incurred by delegating this work to human agents are high. For this reason, an automatic system able to detect abuse in user-generated content is of great interest. There are a number of ways to tackle this problem, but the most commonly seen in practice are word filtering or regular expression matching. The main limitations are their vulnerability to intentional obfuscation on the part of the users, and their context-insensitive nature. Moreover, they are language-dependent and may require appropriate corpora for training. In this paper, we propose a system for automatic abuse detection that completely disregards message content. We first extract a conversational network from raw chat logs and characterize it through topological measures. We then use these as features to train a classifier on our abuse detection task. We thoroughly assess our system on a dataset of user comments originating from a French Massively Multiplayer Online Game. We identify the most appropriate network extraction parameters and discuss the discriminative power of our features, relatively to their topological and temporal nature. Our method reaches an F-measure of 83.89 when using the full feature set, improving on existing approaches. With a selection of the most discriminative features, we dramatically cut computing time while retaining most of the performance (82.65).
http://arxiv.org/abs/1901.11281
Cross-modal retrieval methods have been significantly improved in last years with the use of deep neural networks and large-scale annotated datasets such as ImageNet and Places. However, collecting and annotating such datasets requires a tremendous amount of human effort and, besides, their annotations are usually limited to discrete sets of popular visual classes that may not be representative of the richer semantics found on large-scale cross-modal retrieval datasets. In this paper, we present a self-supervised cross-modal retrieval framework that leverages as training data the correlations between images and text on the entire set of Wikipedia articles. Our method consists in training a CNN to predict: (1) the semantic context of the article in which an image is more probable to appear as an illustration (global context), and (2) the semantic context of its caption (local context). Our experiments demonstrate that the proposed method is not only capable of learning discriminative visual representations for solving vision tasks like image classification and object detection, but that the learned representations are better for cross-modal retrieval when compared to supervised pre-training of the network on the ImageNet dataset.
http://arxiv.org/abs/1902.00378
Recent work in deep learning has shown that the artificial neural networks are vulnerable to adversarial attacks, where a very small perturbation of the inputs can drastically alter the classification result. In this work, we propose a generalized likelihood ratio method capable of training the artificial neural networks with some biological brain-like mechanisms,.e.g., (a) learning by the loss value, (b) learning via neurons with discontinuous activation and loss functions. The traditional back propagation method cannot train the artificial neural networks with aforementioned brain-like learning mechanisms. Numerical results show that various artificial neural networks trained by the new method can significantly improve the defensiveness against the adversarial attacks. Code is available: \url{https://github.com/LX-doctorAI/GLR_ADV} .
http://arxiv.org/abs/1902.00358
Convolutional neural networks have been widely used in content-based image retrieval. To better deal with large-scale data, the deep hashing model is proposed as an effective method, which maps an image to a binary code that can be used for hashing search. However, most existing deep hashing models only utilize fine-level semantic labels or convert them to similar/dissimilar labels for training. The natural semantic hierarchy structures are ignored in the training stage of the deep hashing model. In this paper, we present an effective algorithm to train a deep hashing model that can preserve a semantic hierarchy structure for large-scale image retrieval. Experiments on two datasets show that our method improves the fine-level retrieval performance. Meanwhile, our model achieves state-of-the-art results in terms of hierarchical retrieval.
http://arxiv.org/abs/1901.11259
Unlike holography, fluorescence microscopy lacks an image propagation and time-reversal framework, which necessitates scanning of fluorescent objects to obtain 3D images. We demonstrate that a neural network can inherently learn the physical laws governing fluorescence wave propagation and time-reversal to enable 3D imaging of fluorescent samples using a single 2D image, without mechanical scanning, additional hardware, or a trade-off of resolution or speed. Using this data-driven framework, we increased the depth-of-field of a microscope by 20-fold, imaged Caenorhabditis elegans neurons in 3D using a single fluorescence image, and digitally propagated fluorescence images onto user-defined 3D surfaces, also correcting various aberrations. Furthermore, this learning-based approach cross-connects different imaging modalities, permitting 3D propagation of a wide-field fluorescence image to match confocal microscopy images acquired at different sample planes.
http://arxiv.org/abs/1901.11252
In this paper we demonstrate the effectiveness of a well trained U-Net in the context of the BraTS 2018 challenge. This endeavour is particularly interesting given that researchers are currently besting each other with architectural modifications that are intended to improve the segmentation performance. We instead focus on the training process arguing that a well trained U-Net is hard to beat. Our baseline U-Net, which has only minor modifications and is trained with a large patch size and a Dice loss function indeed achieved competitive Dice scores on the BraTS2018 validation data. By incorporating additional measures such as region based training, additional training data, a simple postprocessing technique and a combination of loss functions, we obtain Dice scores of 77.88, 87.81 and 80.62, and Hausdorff Distances (95th percentile) of 2.90, 6.03 and 5.08 for the enhancing tumor, whole tumor and tumor core, respectively on the test data. This setup achieved rank two in BraTS2018, with more than 60 teams participating in the challenge.
http://arxiv.org/abs/1809.10483
This paper reports on the 2018 PIRM challenge on perceptual super-resolution (SR), held in conjunction with the Perceptual Image Restoration and Manipulation (PIRM) workshop at ECCV 2018. In contrast to previous SR challenges, our evaluation methodology jointly quantifies accuracy and perceptual quality, therefore enabling perceptual-driven methods to compete alongside algorithms that target PSNR maximization. Twenty-one participating teams introduced algorithms which well-improved upon the existing state-of-the-art methods in perceptual SR, as confirmed by a human opinion study. We also analyze popular image quality measures and draw conclusions regarding which of them correlates best with human opinion scores. We conclude with an analysis of the current trends in perceptual SR, as reflected from the leading submissions.
http://arxiv.org/abs/1809.07517
Recovering high-quality images from limited sensory data is a challenging computer vision problem that has received significant attention in recent years. In particular, solutions based on deep learning, ranging from autoencoders to generative models, have been especially effective. However, comparatively little work has centered on the robustness of such reconstructions in terms of the generation of realistic image artifacts (known as hallucinations) and quantifying uncertainty. In this work, we develop experimental methods to address these concerns, utilizing a variational autoencoder-based generative adversarial network (VAE-GAN) as a probabilistic image recovery algorithm. We evaluate the model’s output distribution statistically by exploring the variance, bias, and error associated with generated reconstructions. Furthermore, we perform eigen analysis by examining the Jacobians of outputs with respect to the aliased inputs to more accurately determine which input components can be responsible for deteriorated output quality. Experiments were carried out using a dataset of Knee MRI images, and our results indicate factors such as sampling rate, acquisition model, and loss function impact the model’s robustness. We also conclude that a wise choice of hyperparameters can lead to the robust recovery of MRI images.
http://arxiv.org/abs/1901.11228
A reliable controller is critical and essential for the execution of safe and smooth maneuvers of an autonomous vehicle.The controller must be robust to external disturbances, such as road surface, weather, and wind conditions, and so on.It also needs to deal with the internal parametric variations of vehicle sub-systems, including power-train efficiency, measurement errors, time delay,so on.Moreover, as in most production vehicles, the low-control commands for the engine, brake, and steering systems are delivered through separate electronic control units.These aforementioned factors introduce opaque and ineffectiveness issues in controller performance.In this paper, we design a feed-forward compensate process via a data-driven method to model and further optimize the controller performance.We apply the principal component analysis to the extraction of most influential features.Subsequently,we adopt a time delay neural network and include the accuracy of the predicted error in a future time horizon.Utilizing the predicted error,we then design a feed-forward compensate process to improve the control performance.Finally,we demonstrate the effectiveness of the proposed feed-forward compensate process in simulation scenarios.
http://arxiv.org/abs/1901.11212
A fundamental question lies in almost every application of deep neural networks: what is the optimal neural architecture given a specific dataset? Recently, several Neural Architecture Search (NAS) frameworks have been developed that use reinforcement learning and evolutionary algorithm to search for the solution. However, most of them take a long time to find the optimal architecture due to the huge search space and the lengthy training process needed to evaluate each candidate. In addition, most of them aim at accuracy only and do not take into consideration the hardware that will be used to implement the architecture. This will potentially lead to excessive latencies beyond specifications, rendering the resulting architectures useless. To address both issues, in this paper we use Field Programmable Gate Arrays (FPGAs) as a vehicle to present a novel hardware-aware NAS framework, namely FNAS, which will provide an optimal neural architecture with latency guaranteed to meet the specification. In addition, with a performance abstraction model to analyze the latency of neural architectures without training, our framework can quickly prune architectures that do not satisfy the specification, leading to higher efficiency. Experimental results on common data set such as ImageNet show that in the cases where the state-of-the-art generates architectures with latencies 7.81x longer than the specification, those from FNAS can meet the specs with less than 1% accuracy loss. Moreover, FNAS also achieves up to 11.13x speedup for the search process. To the best of the authors’ knowledge, this is the very first hardware aware NAS.
https://arxiv.org/abs/1901.11211
Deep learning has shown promise to augment radiologists and improve the standard of care globally. Two main issues that complicate deploying these systems are patient privacy and scaling to the global population. To deploy a system at scale with minimal computational cost while preserving privacy we present a web delivered (but locally run) system for diagnosing chest X-Rays. Code is delivered via a URL to a web browser (including cell phones) but the patient data remains on the users machine and all processing occurs locally. The system is designed to be used as a reference where a user can process an image to confirm or aid in their diagnosis. The system contains three main components: out-of-distribution detection, disease prediction, and prediction explanation. The system open source and freely available here: https://mlmed.org/tools/xray/
http://arxiv.org/abs/1901.11210
Object detection in videos has drawn increasing attention recently since it is more important in real scenarios. Most of the deep learning methods for video analysis use convolutional neural networks designed for image-wise parsing in a video stream. But they usually ignore the fact that a video is generally stored and transmitted in a compressed data format. In this paper, we propose a fast object detection model that incorporates light-weight motion-aided memory network (MMNet), which can be directly used for H.264 compressed video. MMNet has two major advantages: 1) For a group of successive pictures (GOP) in a compressed video stream, it runs the heavy computational network for I-frames, i.e. a few reference frames in videos, while a light-weight memory network is designed to generate features for prediction frames called P-frames; 2) Unlike establishing an additional network to explicitly model motion among frames, we directly take full advantage of both motion vectors and residual errors that are all encoded in a compressed video. Such signals maintain spatial variations and are freely available. To our best knowledge, the MMNet is the first work that explores a convolutional detector on a compressed video and a motion-based memory in order to achieve significant speedup. Our model is evaluated on the large-scale ImageNet VID dataset, and the results show that it is about 3x times faster than single image detector R-FCN and 10x times faster than high performance detectors like FGFA and MANet.
https://arxiv.org/abs/1811.11057
In this paper, we propose a lossless data hiding scheme in JPEG images. After quantified DCT transform, coefficients have characteristics that distribution in high frequencies is relatively sparse and absolute values are small. To improve encoding efficiency, we put forward an encoding algorithm that searches for a high frequency as terminate point and recode the coefficients above, so spare space is reserved to embed secret data and appended data with no file expansion. Receiver can obtain terminate point through data analysis, extract additional data and recover original JPEG images lossless. Experimental results show that the proposed method has a larger capacity than state-of-the-art works.
http://arxiv.org/abs/1901.11203
We present EDA: easy data augmentation techniques for boosting performance on text classification tasks. EDA consists of four simple but powerful operations: synonym replacement, random insertion, random swap, and random deletion. On five text classification tasks, we show that EDA improves performance for both convolutional and recurrent neural networks. EDA demonstrates particularly strong results for smaller datasets; on average, across five datasets, training with EDA while using only 50% of the available training set achieved the same accuracy as normal training with all available data. We also performed extensive ablation studies and suggest parameters for practical use.
http://arxiv.org/abs/1901.11196