We show that standard ResNet architectures can be made invertible, allowing the same model to be used for classification, density estimation, and generation. Typically, enforcing invertibility requires partitioning dimensions or restricting network architectures. In contrast, our approach only requires adding a simple normalization step during training, already available in standard frameworks. Invertible ResNets define a generative model which can be trained by maximum likelihood on unlabeled data. To compute likelihoods, we introduce a tractable approximation to the Jacobian log-determinant of a residual block. Our empirical evaluation shows that invertible ResNets perform competitively with both state-of-the-art image classifiers and flow-based generative models, something that has not been previously achieved with a single architecture.
http://arxiv.org/abs/1811.00995
With the rise of the Internet, there is a growing need to build intelligent systems that are capable of efficiently dealing with early risk detection (ERD) problems on social media, such as early depression detection, early rumor detection or identification of sexual predators. These systems, nowadays mostly based on machine learning techniques, must be able to deal with data streams since users provide their data over time. In addition, these systems must be able to decide when the processed data is sufficient to actually classify users. Moreover, since ERD tasks involve risky decisions by which people’s lives could be affected, such systems must also be able to justify their decisions. However, most standard and state-of-the-art supervised machine learning models (such as SVM, MNB, Neural Networks, etc.) are not well suited to deal with this scenario. This is due to the fact that they either act as black boxes or do not support incremental classification/learning. In this paper we introduce SS3, a novel supervised learning model for text classification that naturally supports these aspects. SS3 was designed to be used as a general framework to deal with ERD problems. We evaluated our model on the CLEF’s eRisk2017 pilot task on early depression detection. Most of the 30 contributions submitted to this competition used state-of-the-art methods. Experimental results show that our classifier was able to outperform these models and standard classifiers, despite being less computationally expensive and having the ability to explain its rationale.
http://arxiv.org/abs/1905.08772
In this study we propose a framework to characterize documents based on their semantic flow. The proposed framework encompasses a network-based model that connected sentences based on their semantic similarity. Semantic fields are detected using standard community detection methods. as the story unfolds, transitions between semantic fields are represent in Markov networks, which in turned are characterized via network motifs (subgraphs). Here we show that the proposed framework can be used to classify books according to their style and publication dates. Remarkably, even without a systematic optimization of parameters, philosophy and investigative books were discriminated with an accuracy rate of 92.5%. Because this model captures semantic features of texts, it could be used as an additional feature in traditional network-based models of texts that capture only syntactical/stylistic information, as it is the case of word adjacency (co-occurrence) networks.
http://arxiv.org/abs/1905.07595
For bidirectional joint image-text modeling, we develop variational hetero-encoder (VHE) randomized generative adversarial network (GAN) that integrates a probabilistic text decoder, probabilistic image encoder, and GAN into a coherent end-to-end multi-modality learning framework. VHE randomized GAN (VHE-GAN) encodes an image to decode its associated text, and feeds the variational posterior as the source of randomness into the GAN image generator. We plug three off-the-shelf modules, including a deep topic model, a ladder-structured image encoder, and StackGAN++, into VHE-GAN, which already achieves competitive performance. This further motivates the development of VHE-raster-scan-GAN that generates photo-realistic images in not only a multi-scale low-to-high-resolution manner, but also a hierarchical-semantic coarse-to-fine fashion. By capturing and relating hierarchical semantic and visual concepts with end-to-end training, VHE-raster-scan-GAN achieves state-of-the-art performance in a wide variety of image-text multi-modality learning and generation tasks. PyTorch code is provided.
http://arxiv.org/abs/1905.08622
Recently, pre-trained models have been the dominant paradigm in natural language processing. They achieved remarkable state-of-the-art performance across a wide range of related tasks, such as textual entailment, natural language inference, question answering, etc. BERT, proposed by Devlin et.al., has achieved a better marked result in GLUE leaderboard with a deep transformer architecture. Despite its soaring popularity, however, BERT has not yet been applied to answer selection. This task is different from others with a few nuances: first, modeling the relevance and correctness of candidates matters compared to semantic relatedness and syntactic structure; second, the length of an answer may be different from other candidates and questions. In this paper. we are the first to explore the performance of fine-tuning BERT for answer selection. We achieved STOA results across five popular datasets, demonstrating the success of pre-trained models in this task.
http://arxiv.org/abs/1905.07588
Data size is the bottleneck for developing deep saliency models, because collecting eye-movement data is very time consuming and expensive. Most of current studies on human attention and saliency modeling have used high quality stereotype stimuli. In real world, however, captured images undergo various types of transformations. Can we use these transformations to augment existing saliency datasets? Here, we first create a novel saliency dataset including fixations of 10 observers over 1900 images degraded by 19 types of transformations. Second, by analyzing eye movements, we find that observers look at different locations over transformed versus original images. Third, we utilize the new data over transformed images, called data augmentation transformation (DAT), to train deep saliency models. We find that label preserving DATs with negligible impact on human gaze boost saliency prediction, whereas some other DATs that severely impact human gaze degrade the performance. These label preserving valid augmentation transformations provide a solution to enlarge existing saliency datasets. Finally, we introduce a novel saliency model based on generative adversarial network (dubbed GazeGAN). A modified UNet is proposed as the generator of the GazeGAN, which combines classic skip connections with a novel center-surround connection (CSC), in order to leverage multi level features. We also propose a histogram loss based on Alternative Chi Square Distance (ACS HistLoss) to refine the saliency map in terms of luminance distribution. Extensive experiments and comparisons over 3 datasets indicate that GazeGAN achieves the best performance in terms of popular saliency evaluation metrics, and is more robust to various perturbations. Our code and data are available at: https://github.com/CZHQuality/Sal-CFS-GAN.
http://arxiv.org/abs/1905.06803
Automatic hashtag annotation plays an important role in content understanding for microblog posts. To date, progress made in this field has been restricted to phrase selection from limited candidates, or word-level hashtag discovery using topic models. Different from previous work considering hashtags to be inseparable, our work is the first effort to annotate hashtags with a novel sequence generation framework via viewing the hashtag as a short sequence of words. Moreover, to address the data sparsity issue in processing short microblog posts, we propose to jointly model the target posts and the conversation contexts initiated by them with bidirectional attention. Extensive experimental results on two large-scale datasets, newly collected from English Twitter and Chinese Weibo, show that our model significantly outperforms state-of-the-art models based on classification. Further studies demonstrate our ability to effectively generate rare and even unseen hashtags, which is however not possible for most existing methods.
http://arxiv.org/abs/1905.07584
Our work is a simple extension of the paper “Exploration by Random Network Distillation”. More in detail, we show how to efficiently combine Intrinsic Rewards with Experience Replay in order to achieve more efficient and robust exploration (with respect to PPO/RND) and consequently better results in terms of agent performances and sample efficiency. We are able to do it by using a new technique named Prioritized Oversampled Experience Replay (POER), that has been built upon the definition of what is the important experience useful to replay. Finally, we evaluate our technique on the famous Atari game Montezuma’s Revenge and some other hard exploration Atari games.
http://arxiv.org/abs/1905.07579
We derive a new asymptotic expansion for the global excess risk of a local-$k$-nearest neighbour classifier, where the choice of $k$ may depend upon the test point. This expansion elucidates conditions under which the dominant contribution to the excess risk comes from the decision boundary of the optimal Bayes classifier, but we also show that if these conditions are not satisfied, then the dominant contribution may arise from the tails of the marginal distribution of the features. Moreover, we prove that, provided the $d$-dimensional marginal distribution of the features has a finite $\rho$th moment for some $\rho > 4$ (as well as other regularity conditions), a local choice of $k$ can yield a rate of convergence of the excess risk of $O(n^{-4/(d+4)})$, where $n$ is the sample size, whereas for the standard $k$-nearest neighbour classifier, our theory would require $d \geq 5$ and $\rho > 4d/(d-4)$ finite moments to achieve this rate. These results motivate a new $k$-nearest neighbour classifier for semi-supervised learning problems, where the unlabelled data are used to obtain an estimate of the marginal feature density, and fewer neighbours are used for classification when this density estimate is small. Our worst-case rates are complemented by a minimax lower bound, which reveals that the local, semi-supervised $k$-nearest neighbour classifier attains the minimax optimal rate over our classes for the excess risk, up to a subpolynomial factor in $n$. These theoretical improvements over the standard $k$-nearest neighbour classifier are also illustrated through a simulation study.
http://arxiv.org/abs/1704.00642
Human thinking requires the brain to understand the meaning of language expression and to properly organize the thoughts flow using the language. However, current natural language processing models are primarily limited in the word probability estimation. Here, we proposed a Language guided imagination (LGI) network to incrementally learn the meaning and usage of numerous words and syntaxes, aiming to form a human-like machine thinking process. LGI contains three subsystems: (1) vision system that contains an encoder to disentangle the input or imagined scenarios into abstract population representations, and an imagination decoder to reconstruct imagined scenario from higher level representations; (2) Language system, that contains a binarizer to transfer symbol texts into binary vectors, an IPS (mimicking the human IntraParietal Sulcus, implemented by an LSTM) to extract the quantity information from the input texts, and a textizer to convert binary vectors into text symbols; (3) a PFC (mimicking the human PreFrontal Cortex, implemented by an LSTM) to combine inputs of both language and vision representations, and predict text symbols and manipulated images accordingly. LGI has incrementally learned eight different syntaxes (or tasks), with which a machine thinking loop has been formed and validated by the proper interaction between language and vision system. The paper provides a new architecture to let the machine learn, understand and use language in a human-like way that could ultimately enable a machine to construct fictitious ‘mental’ scenario and possess intelligence.
http://arxiv.org/abs/1905.07562
High Definition Astrometry (0.1 - 1.0 micro-arcseconds) will open a new window into neighboring planetary systems. For the first time, the realm of temperate terrestrial worlds will be explored. This includes Earth Analogs, thereby allowing the value of eta-Earth to be directly determined, without resort to extrapolation. High Definition Astrometry will provide a means to confirm the existence of Radial Velocity (RV) planets while, at the same time, measuring true mass, along with orbit inclination and radius, i.e., system architecture. Planetary systems not amenable to RV search, such as those in a “face-on” orientation, will be surveyed for the first time. Extreme Precision Astrometry is not only useful, but is essential to the future of exoplanet research.
https://arxiv.org/abs/1905.07556
Many computer vision applications require solving multiple tasks in real-time. A neural network can be trained to solve multiple tasks simultaneously using `multi-task learning’. This saves computation at inference time as only a single network needs to be evaluated. Unfortunately, this often leads to inferior overall performance as task objectives compete, which consequently poses the question: which tasks should and should not be learned together in one network when employing multi-task learning? We systematically study task cooperation and competition and propose a framework for assigning tasks to a few neural networks such that cooperating tasks are computed by the same neural network, while competing tasks are computed by different networks. Our framework offers a time-accuracy trade-off and can produce better accuracy using less inference time than not only a single large multi-task neural network but also many single-task networks.
http://arxiv.org/abs/1905.07553
There has been tremendous research progress in estimating the depth of a scene from a monocular camera image. Existing methods for single-image depth prediction are exclusively based on deep neural networks, and their training can be unsupervised using stereo image pairs, supervised using LiDAR point clouds, or semi-supervised using both stereo and LiDAR. In general, semi-supervised training is preferred as it does not suffer from the weaknesses of either supervised training, resulting from the difference in the cameras and the LiDARs field of view, or unsupervised training, resulting from the poor depth accuracy that can be recovered from a stereo pair. In this paper, we present our research in single image depth prediction using semi-supervised training that outperforms the state-of-the-art. We achieve this through a loss function that explicitly exploits left-right consistency in a stereo reconstruction, which has not been adopted in previous semi-supervised training. In addition, we describe the correct use of ground truth depth derived from LiDAR that can significantly reduce prediction error. The performance of our depth prediction model is evaluated on popular datasets, and the importance of each aspect of our semi-supervised training approach is demonstrated through experimental results. Our deep neural network model has been made publicly available.
http://arxiv.org/abs/1905.07542
Architectures obtained by Neural Architecture Search (NAS) have achieved highly competitive performance in various computer vision tasks. However, the prohibitive computation demand of forward-backward propagation in deep neural networks and searching algorithms makes it difficult to apply NAS in practice. In this paper, we propose a Multinomial Distribution Learning for extremely effective NAS, which considers the search space as a joint multinomial distribution, i.e., the operation between two nodes is sampled from this distribution, and the optimal network structure is obtained by the operations with the most likely probability in this distribution. Therefore, NAS can be transformed to a multinomial distribution learning problem, i.e., the distribution is optimized to have high expectation of the performance. Besides, a hypothesis that the performance ranking is consistent in every training epoch is proposed and demonstrated to further accelerate the learning process. Experiments on CIFAR-10 and ImageNet demonstrate the effectiveness of our method. On CIFAR-10, the structure searched by our method achieves 2.4\% test error, while being 6.0 $\times$ (only 4 GPU hours on GTX1080Ti) faster compared with state-of-the-art NAS algorithms. On ImageNet, our model achieves 75.2\% top-1 accuracy under MobileNet settings (MobileNet V1/V2), while being 1.2$\times$ faster with measured GPU latency. Test code is available at https://github.com/tanglang96/MDENAS
http://arxiv.org/abs/1905.07529
Near-range portrait photographs often contain perspective distortion artifacts that bias human perception and challenge both facial recognition and reconstruction techniques. We present the first deep learning based approach to remove such artifacts from unconstrained portraits. In contrast to the previous state-of-the-art approach, our method handles even portraits with extreme perspective distortion, as we avoid the inaccurate and error-prone step of first fitting a 3D face model. Instead, we predict a distortion correction flow map that encodes a per-pixel displacement that removes distortion artifacts when applied to the input image. Our method also automatically infers missing facial features, i.e. occluded ears caused by strong perspective distortion, with coherent details. We demonstrate that our approach significantly outperforms the previous state-of-the-art both qualitatively and quantitatively, particularly for portraits with extreme perspective distortion or facial expressions. We further show that our technique benefits a number of fundamental tasks, significantly improving the accuracy of both face recognition and 3D reconstruction and enables a novel camera calibration technique from a single portrait. Moreover, we also build the first perspective portrait database with a large diversity in identities, expression and poses, which will benefit the related research in this area.
http://arxiv.org/abs/1905.07515
We propose SplitNet, a method for decoupling visual perception and policy learning. By incorporating auxiliary tasks and selective learning of portions of the model, we explicitly decompose the learning objectives for visual navigation into perceiving the world and acting on that perception. We show dramatic improvements over baseline models on transferring between simulators, an encouraging step towards Sim2Real. Additionally, SplitNet generalizes better to unseen environments from the same simulator and transfers faster and more effectively to novel embodied navigation tasks. Further, given only a small sample from a target domain, SplitNet can match the performance of traditional end-to-end pipelines which receive the entire dataset. Code and video are available at https://github.com/facebookresearch/splitnet and https://youtu.be/TJkZcsD2vrc
http://arxiv.org/abs/1905.07512
Topic models are typically evaluated with respect to the global topic distributions that they generate, using metrics such as coherence, but without regard to local (token-level) topic assignments. Token-level assignments are important for downstream tasks such as classification. Even recent models, which aim to improve the quality of these token-level topic assignments, have been evaluated only with respect to global metrics. We propose a task designed to elicit human judgments of token-level topic assignments. We use a variety of topic model types and parameters and discover that global metrics agree poorly with human assignments. Since human evaluation is expensive we propose a variety of automated metrics to evaluate topic models at a local level. Finally, we correlate our proposed metrics with human judgments from the task on several datasets. We show that an evaluation based on the percent of topic switches correlates most strongly with human judgment of local topic quality. We suggest that this new metric, which we call consistency, be adopted alongside global metrics such as topic coherence when evaluating new topic models.
http://arxiv.org/abs/1905.13126
Cross-referencing, which links passages of text to other related passages, can be a valuable study aid for facilitating comprehension of a text. However, cross-referencing requires first, a comprehensive thematic knowledge of the entire corpus, and second, a focused search through the corpus specifically to find such useful connections. Due to this, cross-reference resources are prohibitively expensive and exist only for the most well-studied texts (e.g. religious texts). We develop a topic-based system for automatically producing candidate cross-references which can be easily verified by human annotators. Our system utilizes fine-grained topic modeling with thousands of highly nuanced and specific topics to identify verse pairs which are topically related. We demonstrate that our system can be cost effective compared to having annotators acquire the expertise necessary to produce cross-reference resources unaided.
http://arxiv.org/abs/1905.07508
Deep learning has achieved remarkable results in 3D shape analysis by learning global shape features from the pixel-level over multiple views. Previous methods, however, compute low-level features for entire views without considering part-level information. In contrast, we propose a deep neural network, called Parts4Feature, to learn 3D global features from part-level information in multiple views. We introduce a novel definition of generally semantic parts, which Parts4Feature learns to detect in multiple views from different 3D shape segmentation benchmarks. A key idea of our architecture is that it transfers the ability to detect semantically meaningful parts in multiple views to learn 3D global features. Parts4Feature achieves this by combining a local part detection branch and a global feature learning branch with a shared region proposal module. The global feature learning branch aggregates the detected parts in terms of learned part patterns with a novel multi-attention mechanism, while the region proposal module enables locally and globally discriminative information to be promoted by each other. We demonstrate that Parts4Feature outperforms the state-of-the-art under three large-scale 3D shape benchmarks.
http://arxiv.org/abs/1905.07506
This paper considers the problem of rearrangement planning, i.e finding a sequence of manipulation actions that displace multiple objects from an initial configuration to a given goal configuration. Rearrangement is a critical skill for robots so that they can effectively operate in confined spaces that contain clutter. Examples of tasks that require rearrangement include packing objects inside a bin, wherein objects need to lay according to a predefined pattern. In tight bins, collision-free grasps are often unavailable. Nonprehensile actions, such as pushing and sliding, are preferred because they can be performed using minimalistic end-effectors that can easily be inserted in the bin. Rearrangement with nonprehensile actions is a challenging problem as it requires reasoning about object interactions in a combinatorially large configuration space of multiple objects. This work revisits several existing rearrangement planning techniques and introduces a new one that exploits nested nonprehensile actions by pushing several similar objects simultaneously along the same path, which removes the need to rearrange each object individually. Experiments in simulation and using a real Kuka robotic arm show the ability of the proposed approach to solve difficult rearrangement tasks while reducing the length of the end-effector’s trajectories.
http://arxiv.org/abs/1905.07505
Recent advances, such as GPT and BERT, have shown success in incorporating a pre-trained transformer language model and fine-tuning operation to improve downstream NLP systems. However, this framework still has some fundamental problems in effectively incorporating supervised knowledge from other related tasks. In this study, we investigate a transferable BERT (TransBERT) training framework, which can transfer not only general language knowledge from large-scale unlabeled data but also specific kinds of knowledge from various semantically related supervised tasks, for a target task. Particularly, we propose utilizing three kinds of transfer tasks, including natural language inference, sentiment classification, and next action prediction, to further train BERT based on a pre-trained model. This enables the model to get a better initialization for the target task. We take story ending prediction as the target task to conduct experiments. The final result, an accuracy of 91.8%, dramatically outperforms previous state-of-the-art baseline methods. Several comparative experiments give some helpful suggestions on how to select transfer tasks. Error analysis shows what are the strength and weakness of BERT-based models for story ending prediction.
http://arxiv.org/abs/1905.07504
Learning global features by aggregating information over multiple views has been shown to be effective for 3D shape analysis. For view aggregation in deep learning models, pooling has been applied extensively. However, pooling leads to a loss of the content within views, and the spatial relationship among views, which limits the discriminability of learned features. We propose 3DViewGraph to resolve this issue, which learns 3D global features by more effectively aggregating unordered views with attention. Specifically, unordered views taken around a shape are regarded as view nodes on a view graph. 3DViewGraph first learns a novel latent semantic mapping to project low-level view features into meaningful latent semantic embeddings in a lower dimensional space, which is spanned by latent semantic patterns. Then, the content and spatial information of each pair of view nodes are encoded by a novel spatial pattern correlation, where the correlation is computed among latent semantic patterns. Finally, all spatial pattern correlations are integrated with attention weights learned by a novel attention mechanism. This further increases the discriminability of learned features by highlighting the unordered view nodes with distinctive characteristics and depressing the ones with appearance ambiguity. We show that 3DViewGraph outperforms state-of-the-art methods under three large-scale benchmarks.
http://arxiv.org/abs/1905.07503
This research strives for natural language moment retrieval in long, untrimmed video streams. The problem is not trivial especially when a video contains multiple moments of interests and the language describes complex temporal dependencies, which often happens in real scenarios. We identify two crucial challenges: semantic misalignment and structural misalignment. However, existing approaches treat different moments separately and do not explicitly model complex moment-wise temporal relations. In this paper, we present Moment Alignment Network (MAN), a novel framework that unifies the candidate moment encoding and temporal structural reasoning in a single-shot feed-forward network. MAN naturally assigns candidate moment representations aligned with language semantics over different temporal locations and scales. Most importantly, we propose to explicitly model moment-wise temporal relations as a structured graph and devise an iterative graph adjustment network to jointly learn the best structure in an end-to-end manner. We evaluate the proposed approach on two challenging public benchmarks DiDeMo and Charades-STA, where our MAN significantly outperforms the state-of-the-art by a large margin.
http://arxiv.org/abs/1812.00087
State-of-the-art atmospheric turbulence image restoration methods utilize standard image processing tools such as optical flow, lucky region and blind deconvolution to restore the images. While promising results have been reported over the past decade, many of the methods are agnostic to the physical model that generates the distortion. In this paper, we revisit the turbulence restoration problem by analyzing the reference frame generation and the blind deconvolution steps in a typical restoration pipeline. By leveraging tools in large deviation theory, we rigorously prove the minimum number of frames required to generate a reliable reference for both static and dynamic scenes. We discuss how a turbulence agnostic model can lead to potential flaws, and how to configure a simple spatial-temporal non-local weighted averaging method to generate references. For blind deconvolution, we present a new data-driven prior by analyzing the distributions of the point spread functions. We demonstrate how a simple prior can outperform state-of-the-art blind deconvolution methods.
http://arxiv.org/abs/1905.07498
Speech separation has been studied widely for single-channel close-talk recordings over the past few years; developed solutions are mostly in frequency-domain. Recently, a raw audio waveform separation network (TasNet) introduced for single-channel data, with achieving high Si-SNR (scale-invariant source-to-noise ratio) and SDR (source-to-distortion ratio) comparing against the state-of-the-art solution in frequency-domain. In this study, we incorporate effective components of TasNet into a frequency-domain separation method. We compare both for alternative scenarios. We introduce a solution for directly optimizing the separation criterion in frequency-domain networks. In addition to speech separation objective and subjective measurements, we evaluate the separation performance on a speech recognition task as well. We study the speech separation problem for far-filed data (more similar to naturalistic audio streams) and develop multi-channel solutions for both frequency and time-domain separators with utilizing spectral, spatial and speaker location information. For our experiments, we simulated multi-channel spatialized reverberate WSJ0-2mix dataset. Our experimental results show that spectrogram separation can achieve competitive performance with better network design. With multi-channel framework as well, we can obtain relatively up to +35.5% and +46% improvement in terms of WER and SDR, respectively.
http://arxiv.org/abs/1905.07497
In this paper, we reformulated the spell correction problem as a machine translation task under the encoder-decoder framework. This reformulation enabled us to use a single model for solving the problem that is traditionally formulated as learning a language model and an error model. This model employs multi-layer recurrent neural networks as an encoder and a decoder. We demonstrate the effectiveness of this model using an internal dataset, where the training data is automatically obtained from user logs. The model offers competitive performance as compared to the state of the art methods but does not require any feature engineering nor hand tuning between models.
http://arxiv.org/abs/1705.07371
In the presented scenario, an autonomous surface vehicle (ASV) equipped with a laser scanner navigates on a inland pathway surrounded and crossed by man-made structures such as bridges and locks. {GPS} receiver present on board experiences signal loss and multipath reflections in situation when the view of the sky is obscured by a bridge or tall buildings. In both cases, a potentially dangerous situation is provoked as the robot has no or inaccurate positioning data. A sensor data processing scheme is proposed where these gaps are smoothly filled in by positioning data generated from scan matching and registration of the laser data. This article shows preliminary results of positioning data improvement during trials in harbor-river environment.
http://arxiv.org/abs/1905.07491
The Plug-and-Play (PnP) ADMM algorithm is a powerful image restoration framework that allows advanced image denoising priors to be integrated into physical forward models to generate high quality image restoration results. However, despite the enormous number of applications and several theoretical studies trying to prove the convergence by leveraging tools in convex analysis, very little is known about why the algorithm is doing so well. The goal of this paper is to fill the gap by discussing the performance of PnP ADMM. By restricting the denoisers to the class of graph filters under a linearity assumption, or more specifically the symmetric smoothing filters, we offer three contributions: (1) We show conditions under which an equivalent maximum-a-posteriori (MAP) optimization exists, (2) we present a geometric interpretation and show that the performance gain is due to an intrinsic pre-denoising characteristic of the PnP prior, (3) we introduce a new analysis technique via the concept of consensus equilibrium, and provide interpretations to problems involving multiple priors.
http://arxiv.org/abs/1809.00020
In this paper, we propose a method to obtain a compact and accurate 3D wireframe representation from a single image by effectively exploiting global structural regularities. Our method trains a convolutional neural network to simultaneously detect salient junctions and straight lines, as well as predict their 3D depth and vanishing points. Compared with the state-of-the-art learning-based wireframe detection methods, our network is much simpler and more unified, leading to better 2D wireframe detection. With global structural priors such as Manhattan assumption, our method further reconstructs a full 3D wireframe model, a compact vector representation suitable for a variety of high-level vision tasks such as AR and CAD. We conduct extensive evaluations on a large synthetic dataset of urban scenes as well as real images. Our code and datasets will be released.
http://arxiv.org/abs/1905.07482
Compressive Learning is an emerging topic that combines signal acquisition via compressive sensing and machine learning to perform inference tasks directly on a small number of measurements. Many data modalities naturally have a multi-dimensional or tensorial format, with each dimension or tensor mode representing different features such as the spatial and temporal information in video sequences or the spatial and spectral information in hyperspectral images. However, in existing compressive learning frameworks, the compressive sensing component utilizes either random or learned linear projection on the vectorized signal to perform signal acquisition, thus discarding the multi-dimensional structure of the signals. In this paper, we propose Multilinear Compressive Learning, a framework that takes into account the tensorial nature of multi-dimensional signals in the acquisition step and builds the subsequent inference model on the structurally sensed measurements. Our theoretical complexity analysis shows that the proposed framework is more efficient compared to its vector-based counterpart in both memory and computation requirement. With extensive experiments, we also empirically show that our Multilinear Compressive Learning framework outperforms the vector-based framework in object classification and face recognition tasks, and scales favorably when the dimensionalities of the original signals increase, making it highly efficient for high-dimensional multi-dimensional signals.
http://arxiv.org/abs/1905.07481
Although nowadays advanced dense image matching (DIM) algorithms are able to produce LiDAR (Light Detection And Ranging) comparable dense point clouds from satellite stereo images, the accuracy and completeness of such point clouds heavily depend on the geometric parameters of the satellite stereo images. The intersection angle between two images are normally seen as the most important one in stereo data acquisition, as the state-of-the-art DIM algorithms work best on narrow baseline (smaller intersection angle) stereos (E.g. Semi-Global Matching regards 15-25 degrees as good intersection angle). This factor is in line with the traditional aerial photogrammetry configuration, as the intersection angle directly relates to the base-high ratio and texture distortion in the parallax direction, thus both affecting the horizontal and vertical accuracy. However, our experiments found that even with very similar (and good) intersection angles, the same DIM algorithm applied on different stereo pairs (of the same area) produced point clouds with dramatically different accuracy as compared to the ground truth LiDAR data. This raises a very practical question that is often asked by practitioners: what factors constitute a good satellite stereo pair, such that it produces accurate and optimal results for mapping purpose? In this work, we provide a comprehensive analysis on this matter by performing stereo matching over 1,000 satellite stereo pairs with different acquisition parameters including their intersection angles, off-nadir angles, sun elevation & azimuth angles, as well as time differences, thus to offer a thorough answer to this question. This work will potentially provide a valuable reference to researchers working on multi-view satellite image reconstruction, as well as industrial practitioners minimizing costs for high-quality large-scale mapping.
http://arxiv.org/abs/1905.07476
This paper presents an automated pipeline for processing multi-view satellite images to 3D digital surface models (DSM). The proposed pipeline performs automated geo-referencing and generates high-quality densely matched point clouds. In particular, a novel approach is developed that fuses multiple depth maps derived by stereo matching to generate high-quality 3D maps. By learning critical configurations of stereo pairs from sample LiDAR data, we rank the image pairs based on the proximity of the results to the sample data. Multiple depth maps derived from individual image pairs are fused with an adaptive 3D median filter that considers the image spectral similarities. We demonstrate that the proposed adaptive median filter generally delivers better results in general as compared to normal median filter, and achieved an accuracy of improvement of 0.36 meters RMSE in the best case. Results and analysis are introduced in detail.
http://arxiv.org/abs/1905.07475
This paper presents a novel approach to AUV localization, based on a semantic-aided particle filter. Particle filters have been used successfully for robotics localization since many years. Most of the approaches are however based on geometric measurements and geometric information and simulations. In the past years more and more efforts from research goes towards cognitive robotics and the marine domain is not exception. Moving from signal to symbol becomes therefore paramount for more complex applications. This paper presents a contribution in the well-known area of underwater localization, incorporating semantic information. An extension to the standard particle filter approach is presented, based on semantic information of the environment. A comparison with the geometric approach shows the advantages of a semantic layer to successfully perform self-localization.
http://arxiv.org/abs/1905.07470
Health-related data is noisy and stochastic in implying the true physiological states of patients, limiting information contained in single-moment observations for sequential clinical decision making. We model patient-clinician interactions as partially observable Markov decision processes (POMDPs) and optimize sequential treatment based on belief states inferred from history sequence. To facilitate inference, we build a variational generative model and boost state representation with a recurrent neural network (RNN), incorporating an auxiliary loss from sequence auto-encoding. Meanwhile, we optimize a continuous policy of drug levels with an actor-critic method where policy gradients are obtained from a stablized off-policy estimate of advantage function, with the value of belief state backed up by parallel best-first suffix trees. We exploit our methodology in optimizing dosages of vasopressor and intravenous fluid for sepsis patients using a retrospective intensive care dataset and evaluate the learned policy with off-policy policy evaluation (OPPE). The results demonstrate that modelling as POMDPs yields better performance than MDPs, and that incorporating heuristic search improves sample efficiency.
http://arxiv.org/abs/1905.07465
Preventable adverse drug reactions as a result of medical errors present a growing concern in modern medicine. As drug-drug interactions (DDIs) may cause adverse reactions, being able to extracting DDIs from drug labels into machine-readable form is an important effort in effectively deploying drug safety information. The DDI track of TAC 2018 introduces two large hand-annotated test sets for the task of extracting DDIs from structured product labels with linkage to standard terminologies. Herein, we describe our approach to tackling tasks one and two of the DDI track, which corresponds to named entity recognition (NER) and sentence-level relation extraction respectively. Namely, our approach resembles a multi-task learning framework designed to jointly model various sub-tasks including NER and interaction type and outcome prediction. On NER, our system ranked second (among eight teams) at 33.00% and 38.25% F1 on Test Sets 1 and 2 respectively. On relation extraction, our system ranked second (among four teams) at 21.59% and 23.55% on Test Sets 1 and 2 respectively.
http://arxiv.org/abs/1905.07464
Relation extraction (RE) is an indispensable information extraction task in several disciplines. RE models typically assume that named entity recognition (NER) is already performed in a previous step by another independent model. Several recent efforts, under the theme of end-to-end RE, seek to exploit inter-task correlations by modeling both NER and RE tasks jointly. Earlier work in this area commonly reduces the task to a table-filling problem wherein an additional expensive decoding step involving beam search is applied to obtain globally consistent cell labels. In efforts that do not employ table-filling, global optimization in the form of CRFs with Viterbi decoding for the NER component is still necessary for competitive performance. We introduce a novel neural architecture utilizing the table structure, based on repeated applications of 2D convolutions for pooling local dependency and metric-based features, without the need for global optimization. We validate our model on the ADE and CoNLL04 datasets for end-to-end RE and demonstrate $\approx 1\%$ gain (in F-score) over prior best results with training and testing times that are nearly four times faster — the latter highly advantageous for time-sensitive end user applications.
http://arxiv.org/abs/1905.07458
Machine learning has become ubiquitous and a key technology on mining electronic health records (EHRs) for facilitating clinical research and practice. Unsupervised machine learning, as opposed to supervised learning, has shown promise in identifying novel patterns and relations from EHRs without using human created labels. In this paper, we investigate the application of unsupervised machine learning models in discovering latent disease clusters and patient subgroups based on EHRs. We utilized Latent Dirichlet Allocation (LDA), a generative probabilistic model, and proposed a novel model named Poisson Dirichlet Model (PDM), which extends the LDA approach using a Poisson distribution to model patients’ disease diagnoses and to alleviate age and sex factors by considering both observed and expected observations. In the empirical experiments, we evaluated LDA and PDM on three patient cohorts with EHR data retrieved from the Rochester Epidemiology Project (REP), for the discovery of latent disease clusters and patient subgroups. We compared the effectiveness of LDA and PDM in identifying latent disease clusters through the visualization of disease representations learned by two approaches. We also tested the performance of LDA and PDM in differentiating patient subgroups through survival analysis, as well as statistical analysis. The experimental results show that the proposed PDM could effectively identify distinguished disease clusters by alleviating the impact of age and sex, and that LDA could stratify patients into more differentiable subgroups than PDM in terms of p-values. However, the subgroups discovered by PDM might imply the underlying patterns of diseases of greater interest in epidemiology research due to the alleviation of age and sex. Both unsupervised machine learning approaches could be leveraged to discover patient subgroups using EHRs but with different foci.
http://arxiv.org/abs/1905.10309
Standardized evaluation measures have aided in the progress of machine learning approaches in disciplines such as computer vision and machine translation. In this paper, we make the case that robotic learning would also benefit from benchmarking, and present the “REPLAB” platform for benchmarking vision-based manipulation tasks. REPLAB is a reproducible and self-contained hardware stack (robot arm, camera, and workspace) that costs about 2000 USD, occupies a cuboid of size 70x40x60 cm, and permits full assembly within a few hours. Through this low-cost, compact design, REPLAB aims to drive wide participation by lowering the barrier to entry into robotics and to enable easy scaling to many robots. We envision REPLAB as a framework for reproducible research across manipulation tasks, and as a step in this direction, we define a template for a grasping benchmark consisting of a task definition, evaluation protocol, performance measures, and a dataset of 92k grasp attempts. We implement, evaluate, and analyze several previously proposed grasping approaches to establish baselines for this benchmark. Finally, we also implement and evaluate a deep reinforcement learning approach for 3D reaching tasks on our REPLAB platform. Project page with assembly instructions, code, and videos: https://goo.gl/5F9dP4.
http://arxiv.org/abs/1905.07447
Accurate facial expression analysis is an essential step in various clinical applications that involve physical and mental health assessments of older adults (e.g. diagnosis of pain or depression). Although remarkable progress has been achieved toward developing robust facial landmark detection methods, state-of-the-art methods still face many challenges when encountering uncontrolled environments, different ranges of facial expressions, and different demographics of the population. A recent study has revealed that the health status of individuals can also affect the performance of facial landmark detection methods on front views of faces. In this work, we investigate this matter in a much greater context using seven facial landmark detection methods. We perform our evaluation not only on frontal faces but also on profile faces and in various regions of the face. Our results shed light on limitations of the existing methods and challenges of applying these methods in clinical settings by indicating: 1) a significant difference between the performance of state-of-the-art when tested on the profile or frontal faces of individuals with vs. without dementia; 2) insights on the existing bias for all regions of the face; and 3) the presence of this bias despite re-training/fine-tuning with various configurations of six datasets.
http://arxiv.org/abs/1905.07446
Much research work in computer vision is being spent on optimizing existing network architectures to obtain a few more percentage points on benchmarks. Recent AutoML approaches promise to relieve us from this effort. However, they are mainly designed for comparatively small-scale classification tasks. In this work, we show how to use and extend existing AutoML techniques to efficiently optimize large-scale U-Net-like encoder-decoder architectures. In particular, we leverage gradient-based neural architecture search and Bayesian optimization for hyperparameter search. The resulting optimization does not require a large company-scale compute cluster. We show results on disparity estimation that clearly outperform the manually optimized baseline and reach state-of-the-art performance.
http://arxiv.org/abs/1905.07443
This paper presents an algorithm for enumerating biases in word embeddings. The algorithm exposes a large number of offensive associations related to sensitive features such as race and gender on publicly available embeddings, including a supposedly “debiased” embedding. These biases are concerning in light of the widespread use of word embeddings. The associations are identified by geometric patterns in word embeddings that run parallel between people’s names and common lower-case tokens. The algorithm is highly unsupervised: it does not even require the sensitive features to be pre-specified. This is desirable because: (a) many forms of discrimination–such as racial discrimination–are linked to social constructs that may vary depending on the context, rather than to categories with fixed definitions; and (b) it makes it easier to identify biases against intersectional groups, which depend on combinations of sensitive features. The inputs to our algorithm are a list of target tokens, e.g. names, and a word embedding. It outputs a number of Word Embedding Association Tests (WEATs) that capture various biases present in the data. We illustrate the utility of our approach on publicly available word embeddings and lists of names, and evaluate its output using crowdsourcing. We also show how removing names may not remove potential proxy bias.
http://arxiv.org/abs/1812.08769
This work addresses the problem of semantic foggy scene understanding (SFSU). Although extensive research has been performed on image dehazing and on semantic scene understanding with clear-weather images, little attention has been paid to SFSU. Due to the difficulty of collecting and annotating foggy images, we choose to generate synthetic fog on real images that depict clear-weather outdoor scenes, and then leverage these partially synthetic data for SFSU by employing state-of-the-art convolutional neural networks (CNN). In particular, a complete pipeline to add synthetic fog to real, clear-weather images using incomplete depth information is developed. We apply our fog synthesis on the Cityscapes dataset and generate Foggy Cityscapes with 20550 images. SFSU is tackled in two ways: 1) with typical supervised learning, and 2) with a novel type of semi-supervised learning, which combines 1) with an unsupervised supervision transfer from clear-weather images to their synthetic foggy counterparts. In addition, we carefully study the usefulness of image dehazing for SFSU. For evaluation, we present Foggy Driving, a dataset with 101 real-world images depicting foggy driving scenes, which come with ground truth annotations for semantic segmentation and object detection. Extensive experiments show that 1) supervised learning with our synthetic data significantly improves the performance of state-of-the-art CNN for SFSU on Foggy Driving; 2) our semi-supervised learning strategy further improves performance; and 3) image dehazing marginally advances SFSU with our learning strategy. The datasets, models and code are made publicly available.
http://arxiv.org/abs/1708.07819
To help enforce data-protection regulations such as GDPR and detect unauthorized uses of personal data, we develop a new \emph{model auditing} technique that helps users check if their data was used to train a machine learning model. We focus on auditing deep-learning models that generate natural-language text, including word prediction and dialog generation. These models are at the core of popular online services and are often trained on personal data such as users’ messages, searches, chats, and comments. We design and evaluate a black-box auditing method that can detect, with very few queries to a model, if a particular user’s texts were used to train it (among thousands of other users). We empirically show that our method can successfully audit well-generalized models that are not overfitted to the training data. We also analyze how text-generation models memorize word sequences and explain why this memorization makes them amenable to auditing.
http://arxiv.org/abs/1811.00513
Model-agnostic meta-learning (MAML) is a meta-learning technique to train a model on a multitude of learning tasks in a way that primes the model for few-shot learning of new tasks. The MAML algorithm performs well on few-shot learning problems in classification, regression, and fine-tuning of policy gradients in reinforcement learning, but comes with the need for costly hyperparameter tuning for training stability. We address this shortcoming by introducing an extension to MAML, called Alpha MAML, to incorporate an online hyperparameter adaptation scheme that eliminates the need to tune meta-learning and learning rates. Our results with the Omniglot database demonstrate a substantial reduction in the need to tune MAML training hyperparameters and improvement to training stability with less sensitivity to hyperparameter choice.
http://arxiv.org/abs/1905.07435
The problem of coordination without a priori information about the environment is important in robotics. Applications vary from formation control to search and rescue. This paper considers the problem of search by a group of solitary robots: self-interested robots without a priori knowledge about each other, and with restricted communication capacity. When the capacity of robots to communicate is limited, they may obliviously search in overlapping regions (i.e. be subject to interference). Interference hinders robot progress, and strategies have been proposed in the literature to mitigate interference [1], [2]. Interaction of solitary robots has attracted much interest in robotics, but the problem of mitigating interference when time for search is limited remains an important area of research. We propose a coordination strategy based on the method of cellular decomposition [3] where we employ the concept of soft obstacles: a robot considers cells assigned to other robots as obstacles. The performance of the proposed strategy is demonstrated by means of simulation experiments. Simulations indicate the utility of the strategy in situations where a known upper bound on the search time precludes search of the entire environment.
http://arxiv.org/abs/1905.07434
We use Bayesian convolutional neural networks and a novel generative model of Galaxy Zoo volunteer responses to infer posteriors for the visual morphology of galaxies. Bayesian CNN can learn from galaxy images with uncertain labels and then, for previously unlabelled galaxies, predict the probability of each possible label. Our posteriors are well-calibrated (e.g. for predicting bars, we achieve coverage errors of 10.6% within 5 responses and 2.9% within 10 responses) and hence are reliable for practical use. Further, using our posteriors, we apply the active learning strategy BALD to request volunteer responses for the subset of galaxies which, if labelled, would be most informative for training our network. We show that training our Bayesian CNNs using active learning requires up to 35-60% fewer labelled galaxies, depending on the morphological feature being classified. By combining human and machine intelligence, Galaxy Zoo will be able to classify surveys of any conceivable scale on a timescale of weeks, providing massive and detailed morphology catalogues to support research into galaxy evolution.
http://arxiv.org/abs/1905.07424
We present a semi-blind, spatially-variant deconvolution technique aimed at optical microscopy that combines a local estimation step of the point spread function (PSF) and deconvolution using a spatially variant, regularized Richardson-Lucy algorithm. To find the local PSF map in a computationally tractable way, we train a convolutional neural network to perform regression of an optical parametric model on synthetically blurred image patches. We deconvolved both synthetic and experimentally-acquired data, and achieved an improvement of image SNR of 1.00 dB on average, compared to other deconvolution algorithms.
http://arxiv.org/abs/1803.07452
Deep-learning is a cutting edge theory that is being applied to many fields. For vision applications the Convolutional Neural Networks (CNN) are demanding significant accuracy for classification tasks. Numerous hardware accelerators have populated during the last years to improve CPU or GPU based solutions. This technology is commonly prototyped and tested over FPGAs before being considered for ASIC fabrication for mass production. The use of commercial typical cameras (30fps) limits the capabilities of these systems for high speed applications. The use of dynamic vision sensors (DVS) that emulate the behavior of a biological retina is taking an incremental importance to improve this applications due to its nature, where the information is represented by a continuous stream of spikes and the frames to be processed by the CNN are constructed collecting a fixed number of these spikes (called events). The faster an object is, the more events are produced by DVS, so the higher is the equivalent frame rate. Therefore, these DVS utilization allows to compute a frame at the maximum speed a CNN accelerator can offer. In this paper we present a VHDL/HLS description of a pipelined design for FPGA able to collect events from an Address-Event-Representation (AER) DVS retina to obtain a normalized histogram to be used by a particular CNN accelerator, called NullHop. VHDL is used to describe the circuit, and HLS for computation blocks, which are used to perform the normalization of a frame needed for the CNN. Results outperform previous implementations of frames collection and normalization using ARM processors running at 800MHz on a Zynq7100 in both latency and power consumption. A measured 67% speedup factor is presented for a Roshambo CNN real-time experiment running at 160fps peak rate.
http://arxiv.org/abs/1905.07419
The vulnerability to adversarial attacks has been a critical issue for deep neural networks. Addressing this issue requires a reliable way to evaluate the robustness of a network. Recently, several methods have been developed to compute $\textit{robustness quantification}$ for neural networks, namely, certified lower bounds of the minimum adversarial perturbation. Such methods, however, were devised for feed-forward networks, e.g. multi-layer perceptron or convolutional networks. It remains an open problem to quantify robustness for recurrent networks, especially LSTM and GRU. For such networks, there exist additional challenges in computing the robustness quantification, such as handling the inputs at multiple steps and the interaction between gates and states. In this work, we propose $\textit{POPQORN}$ ($\textbf{P}$ropagated-$\textbf{o}$ut$\textbf{p}$ut $\textbf{Q}$uantified R$\textbf{o}$bustness for $\textbf{RN}$Ns), a general algorithm to quantify robustness of RNNs, including vanilla RNNs, LSTMs, and GRUs. We demonstrate its effectiveness on different network architectures and show that the robustness quantification on individual steps can lead to new insights.
http://arxiv.org/abs/1905.07387
Many problems in video understanding require labeling multiple activities occurring concurrently in different parts of a video, including the objects and actors participating in such activities. However, state-of-the-art methods in computer vision focus primarily on tasks such as action classification, action detection, or action segmentation, where typically only one action label needs to be predicted. In this work, we propose a generic approach to classifying one or more nodes of a spatio-temporal graph grounded on spatially localized semantic entities in a video, such as actors and objects. In particular, we combine an attributed spatio-temporal visual graph, which captures visual context and interactions, with an attributed symbolic graph grounded on the semantic label space, which captures relationships between multiple labels. We further propose a neural message passing framework for jointly refining the representations of the nodes and edges of the hybrid visual-symbolic graph. Our framework features a) node-type and edge-type conditioned filters and adaptive graph connectivity, b) a soft-assignment module for connecting visual nodes to symbolic nodes and vice versa, c) a symbolic graph reasoning module that enforces semantic coherence and d) a pooling module for aggregating the refined node and edge representations for downstream classification tasks. We demonstrate the generality of our approach on a variety of tasks, such as temporal subactivity classification and object affordance classification on the CAD-120 dataset and multilabel temporal action localization on the large scale Charades dataset, where we outperform existing deep learning approaches, using only raw RGB frames.
http://arxiv.org/abs/1905.07385