Although considerable attention has been given to neural ranking architectures recently, far less attention has been paid to the term representations that are used as input to these models. In this work, we investigate how two pretrained contextualized language modes (ELMo and BERT) can be utilized for ad-hoc document ranking. Through experiments on TREC benchmarks, we find that several existing neural ranking architectures can benefit from the additional context provided by contextualized language models. Furthermore, we propose a joint approach that incorporates BERT’s classification vector into existing neural models and show that it outperforms state-of-the-art ad-hoc ranking baselines. We also address practical challenges in using these models for ranking, including the maximum input length imposed by BERT and runtime performance impacts of contextualized language models.
http://arxiv.org/abs/1904.07094
We present SIMCO, the first agnostic multi-class object counting approach. SIMCO starts by detecting foreground objects through a novel Mask RCNN-based architecture trained beforehand (just once) on a brand-new synthetic 2D shape dataset, InShape; the idea is to highlight every object resembling a primitive 2D shape (circle, square, rectangle, etc.). Each object detected is described by a low-dimensional embedding, obtained from a novel similarity-based head branch; this latter implements a triplet loss, encouraging similar objects (same 2D shape + color and scale) to map close. Subsequently, SIMCO uses this embedding for clustering, so that different types of objects can emerge and be counted, making SIMCO the very first multi-class unsupervised counter. Experiments show that SIMCO provides state-of-the-art scores on counting benchmarks and that it can also help in many challenging image understanding tasks.
http://arxiv.org/abs/1904.07092
Deep learning-based, single-view depth estimation methods have recently shown highly promising results. However, such methods ignore one of the most important features for determining depth in the human vision system, which is motion. We propose a learning-based, multi-view dense depth map and odometry estimation method that uses Recurrent Neural Networks (RNN) and trains utilizing multi-view image reprojection and forward-backward flow-consistency losses. Our model can be trained in a supervised or even unsupervised mode. It is designed for depth and visual odometry estimation from video where the input frames are temporally correlated. However, it also generalizes to single-view depth estimation. Our method produces superior results to the state-of-the-art approaches for single-view and multi-view learning-based depth estimation on the KITTI driving dataset.
http://arxiv.org/abs/1904.07087
Benefit from the quick development of deep learning techniques, salient object detection has achieved remarkable progresses recently. However, there still exists following two major challenges that hinder its application in embedded devices, low resolution output and heavy model weight. To this end, this paper presents an accurate yet compact deep network for efficient salient object detection. More specifically, given a coarse saliency prediction in the deepest layer, we first employ residual learning to learn side-output residual features for saliency refinement, which can be achieved with very limited convolutional parameters while keep accuracy. Secondly, we further propose reverse attention to guide such side-output residual learning in a top-down manner. By erasing the current predicted salient regions from side-output features, the network can eventually explore the missing object parts and details which results in high resolution and accuracy. Experiments on six benchmark datasets demonstrate that the proposed approach compares favorably against state-of-the-art methods, and with advantages in terms of simplicity, efficiency (45 FPS) and model size (81 MB).
https://arxiv.org/abs/1807.09940
When watching omnidirectional images (ODIs), subjects can access different viewports by moving their heads. Therefore, it is necessary to predict subjects’ head fixations on ODIs. Inspired by generative adversarial imitation learning (GAIL), this paper proposes a novel approach to predict saliency of head fixations on ODIs, named SalGAIL. First, we establish a dataset for attention on ODIs (AOI). In contrast to traditional datasets, our AOI dataset is large-scale, which contains the head fixations of 30 subjects viewing 600 ODIs. Next, we mine our AOI dataset and determine three findings: (1) The consistency of head fixations are consistent among subjects, and it grows alongside the increased subject number; (2) The head fixations exist with a front center bias (FCB); and (3) The magnitude of head movement is similar across subjects. According to these findings, our SalGAIL approach applies deep reinforcement learning (DRL) to predict the head fixations of one subject, in which GAIL learns the reward of DRL, rather than the traditional human-designed reward. Then, multi-stream DRL is developed to yield the head fixations of different subjects, and the saliency map of an ODI is generated via convoluting predicted head fixations. Finally, experiments validate the effectiveness of our approach in predicting saliency maps of ODIs, significantly better than 10 state-of-the-art approaches.
http://arxiv.org/abs/1904.07080
A number of recent studies have started to investigate how speech systems can be trained on untranscribed speech by leveraging accompanying images at training time. Examples of tasks include keyword prediction and within- and across-mode retrieval. Here we consider how such models can be used for query-by-example (QbE) search, the task of retrieving utterances relevant to a given spoken query. We are particularly interested in semantic QbE, where the task is not only to retrieve utterances containing exact instances of the query, but also utterances whose meaning is relevant to the query. We follow a segmental QbE approach where variable-duration speech segments (queries, search utterances) are mapped to fixed-dimensional embedding vectors. We show that a QbE system using an embedding function trained on visually grounded speech data outperforms a purely acoustic QbE system in terms of both exact and semantic retrieval performance.
http://arxiv.org/abs/1904.07078
Physical design process commonly consumes hours to days for large designs, and routing is known as the most critical step. Demands for accurate routing quality prediction raise to a new level to accelerate hardware innovation with advanced technology nodes. This work presents an approach that forecasts the density of all routing channels over the entire floorplan, with features collected up to placement, using conditional GANs. Specifically, forecasting the routing congestion is constructed as an image translation (colorization) problem. The proposed approach is applied to a) placement exploration for minimum congestion, b) constrained placement exploration and c) forecasting congestion in real-time during incremental placement, using eight designs targeting a fixed FPGA architecture.
http://arxiv.org/abs/1904.07077
We investigate unsupervised models that can map a variable-duration speech segment to a fixed-dimensional representation. In settings where unlabelled speech is the only available resource, such acoustic word embeddings can form the basis for “zero-resource” speech search, discovery and indexing systems. Most existing unsupervised embedding methods still use some supervision, such as word or phoneme boundaries. Here we propose the encoder-decoder correspondence autoencoder (EncDec-CAE), which, instead of true word segments, uses automatically discovered segments: an unsupervised term discovery system finds pairs of words of the same unknown type, and the EncDec-CAE is trained to reconstruct one word given the other as input. We compare it to a standard encoder-decoder autoencoder (AE), a variational AE with a prior over its latent embedding, and downsampling. EncDec-CAE outperforms its closest competitor by 24% relative in average precision on two languages in a word discrimination task.
http://arxiv.org/abs/1811.00403
Endoscopy is a routine imaging technique used for both diagnosis and minimally invasive surgical treatment. Artifacts such as motion blur, bubbles, specular reflections, floating objects and pixel saturation impede the visual interpretation and the automated analysis of endoscopy videos. Given the widespread use of endoscopy in different clinical applications, we contend that the robust and reliable identification of such artifacts and the automated restoration of corrupted video frames is a fundamental medical imaging problem. Existing state-of-the-art methods only deal with the detection and restoration of selected artifacts. However, typically endoscopy videos contain numerous artifacts which motivates to establish a comprehensive solution. We propose a fully automatic framework that can: 1) detect and classify six different primary artifacts, 2) provide a quality score for each frame and 3) restore mildly corrupted frames. To detect different artifacts our framework exploits fast multi-scale, single stage convolutional neural network detector. We introduce a quality metric to assess frame quality and predict image restoration success. Generative adversarial networks with carefully chosen regularization are finally used to restore corrupted frames. Our detector yields the highest mean average precision (mAP at 5% threshold) of 49.0 and the lowest computational time of 88 ms allowing for accurate real-time processing. Our restoration models for blind deblurring, saturation correction and inpainting demonstrate significant improvements over previous methods. On a set of 10 test videos we show that our approach preserves an average of 68.7% which is 25% more frames than that retained from the raw videos.
http://arxiv.org/abs/1904.07073
With the increasing use of the Internet and mobile devices, social networks are becoming the most used media to communicate citizens’ ideas and thoughts. This information is very useful to identify communities with common ideas based on what they publish in the network. This paper presents a method to automatically detect city communities based on machine learning techniques applied to a set of tweets from Bogot'a’s citizens. An analysis was performed in a collection of 2,634,176 tweets gathered from Twitter in a period of six months. Results show that the proposed method is an interesting tool to characterize a city population based on a machine learning methods and text analytics.
http://arxiv.org/abs/1904.08926
Synthesising 3D facial motion from speech is a crucial problem manifesting in a multitude of applications such as computer games and movies. Recently proposed methods tackle this problem in controlled conditions of speech. In this paper, we introduce the first methodology for 3D facial motion synthesis from speech captured in arbitrary recording conditions (“in-the-wild”) and independent of the speaker. For our purposes, we captured 4D sequences of people uttering 500 words, contained in the Lip Reading Words (LRW) a publicly available large-scale in-the-wild dataset, and built a set of 3D blendshapes appropriate for speech. We correlate the 3D shape parameters of the speech blendshapes to the LRW audio samples by means of a novel time-warping technique, named Deep Canonical Attentional Warping (DCAW), that can simultaneously learn hierarchical non-linear representations and a warping path in an end-to-end manner. We thoroughly evaluate our proposed methods, and show the ability of a deep learning model to synthesise 3D facial motion in handling different speakers and continuous speech signals in uncontrolled conditions.
http://arxiv.org/abs/1904.07002
The existing Zero-Shot learning (ZSL) methods may suffer from the vague class attributes that are highly overlapped for different classes. Unlike these methods that ignore the discrimination among classes, in this paper, we propose to classify unseen image by rectifying the semantic space guided by the visual space. First, we pre-train a Semantic Rectifying Network (SRN) to rectify semantic space with a semantic loss and a rectifying loss. Then, a Semantic Rectifying Generative Adversarial Network (SR-GAN) is built to generate plausible visual feature of unseen class from both semantic feature and rectified semantic feature. To guarantee the effectiveness of rectified semantic features and synthetic visual features, a pre-reconstruction and a post reconstruction networks are proposed, which keep the consistency between visual feature and semantic feature. Experimental results demonstrate that our approach significantly outperforms the state-of-the-arts on four benchmark datasets.
http://arxiv.org/abs/1904.06996
Ego-motion estimation is a fundamental requirement for most mobile robotic applications. By sensor fusion, we can compensate the deficiencies of stand-alone sensors and provide more reliable estimations. We introduce a tightly coupled lidar-IMU fusion method in this paper. By jointly minimizing the cost derived from lidar and IMU measurements, the lidar-IMU odometry (LIO) can perform well with acceptable drift after long-term experiment, even in challenging cases where the lidar measurements can be degraded. Besides, to obtain more reliable estimations of the lidar poses, a rotation-constrained refinement algorithm (LIO-mapping) is proposed to further align the lidar poses with the global map. The experiment results demonstrate that the proposed method can estimate the poses of the sensor pair at the IMU update rate with high precision, even under fast motion conditions or with insufficient features.
http://arxiv.org/abs/1904.06993
Standard artificial neural networks suffer from the well-known issue of catastrophic forgetting, making continual or lifelong learning difficult for machine learning. In recent years, numerous methods have been proposed for continual learning, but due to differences in evaluation protocols it is difficult to directly compare their performance. To enable more structured comparisons, we describe three continual learning scenarios based on whether at test time task identity is provided and–in case it is not–whether it must be inferred. Any sequence of well-defined tasks can be performed according to each scenario. Using the split and permuted MNIST task protocols, for each scenario we carry out an extensive comparison of recently proposed continual learning methods. We demonstrate substantial differences between the three scenarios in terms of difficulty and in terms of how efficient different methods are. In particular, when task identity must be inferred (i.e., class incremental learning), we find that regularization-based approaches (e.g., elastic weight consolidation) fail and that replaying representations of previous experiences seems required for solving this scenario.
http://arxiv.org/abs/1904.07734
Sequencing large number of candidate disease genes which cause diseases in order to identify the relationship between them is an expensive and time-consuming task. To handle these challenges, different computational approaches have been developed. Based on the observation that genes associated with similar diseases have a higher likelihood of interaction, a large class of these approaches relay on analyzing the topological properties of biological networks. However, the incomplete and noisy nature of biological networks is known as an important challenge in these approaches. In this paper, we propose a two-step framework for disease gene prioritization: (1) construction of a reliable human FLN using sequence information and machine learning techniques, (2) prioritizing the disease gene relations based on the constructed FLN. On our framework, unlike other FLN based frameworks that using FLNs based on integration of various low quality biological data, the sequence of proteins is used as the comprehensive data to construct a reliable initial network. In addition, the physicochemical properties of amino-acids are employed to describe the functionality of proteins. All in all, the proposed approach is evaluated and the results indicate the high efficiency and validity of the FLN in disease gene prioritization.
http://arxiv.org/abs/1904.06973
Implicit discourse relation classification is one of the most challenging and important tasks in discourse parsing, due to the lack of connective as strong linguistic cues. A principle bottleneck to further improvement is the shortage of training data (ca.~16k instances in the PDTB). Shi et al. (2017) proposed to acquire additional data by exploiting connectives in translation: human translators mark discourse relations which are implicit in the source language explicitly in the translation. Using back-translations of such explicitated connectives improves discourse relation parsing performance. This paper addresses the open question of whether the choice of the translation language matters, and whether multiple translations into different languages can be effectively used to improve the quality of the additional data.
http://arxiv.org/abs/1808.10290
We propose a new image denoising algorithm, dubbed as Fully Convolutional Adaptive Image DEnoiser (FC-AIDE), that can learn from offline supervised training set with a fully convolutional neural network as well as adaptively fine-tune the supervised model for each given noisy image. We significantly extend the framework of the recently proposed Neural AIDE, which formulates the denoiser to be context-based pixelwise mappings and utilizes the unbiased estimator of MSE for such denoisers. The two main contributions we make are; 1) implementing a novel fully convolutional architecture that boosts the base supervised model, and 2) introducing regularization methods for the adaptive fine-tuning such that a stronger and more robust adaptivity can be attained. As a result, FC-AIDE is shown to possess many desirable features; it outperforms the recent CNN-based state-of-the-art denoisers on all of the benchmark datasets we tested, and gets particularly strong for various challenging scenarios, e.g., with mismatched image/noise characteristics or with scarce supervised training data.
http://arxiv.org/abs/1807.07569
Gleason grading specified in ISUP 2014 is the clinical standard in staging prostate cancer and the most important part of the treatment decision. However, the grading is subjective and suffers from high intra and inter-user variability. To improve the consistency and objectivity in the grading, we introduced glandular tissue WithOut Basal cells (WOB) as the ground truth. The presence of basal cells is the most accepted biomarker for benign glandular tissue and the absence of basal cells is a strong indicator of acinar prostatic adenocarcinoma, the most common form of prostate cancer. Glandular tissue can objectively be assessed as WOB or not WOB by using specific immunostaining for glandular tissue (Cytokeratin 8/18) and for basal cells (Cytokeratin 5/6 + p63). Even more, WOB allowed us to develop a semi-automated data generation pipeline to speed up the tremendously time consuming and expensive process of annotating whole slide images by pathologists. We generated 295 prostatectomy images exhaustively annotated with WOB. Then we used our Deep Learning Framework, which achieved the $2^{nd}$ best reported score in Camelyon17 Challenge, to train networks for segmenting WOB in needle biopsies. Evaluation of the model on 63 needle biopsies showed promising results which were improved further by finetuning the model on 118 biopsies annotated with WOB, achieving F1-score of 0.80 and Precision-Recall AUC of 0.89 at the pixel-level. Then we compared the performance of the model against 17 biopsies annotated independently by 3 pathologists using only H\&E staining. The comparison demonstrated that the model performed on a par with the pathologists. Finally, the model detected and accurately outlined existing WOB areas in two biopsies incorrectly annotated as totally WOB-free biopsies by three pathologists and in one biopsy by two pathologists.
http://arxiv.org/abs/1904.06969
In this paper, we study dependence of the success rate of adversarial attacks to the Deep Neural Networks on the biomedical image type, control parameters, and image dataset size. With this work, we are going to contribute towards accumulation of experimental results on adversarial attacks for the community dealing with biomedical images. The white-box Projected Gradient Descent attacks were examined based on 8 classification tasks and 13 image datasets containing a total of 605,080 chest X-ray and 317,000 histology images of malignant tumors. We concluded that: (1) An increase of the amplitude of perturbation in generating malicious adversarial images leads to a growth of the fraction of successful attacks for the majority of image types examined in this study. (2) Histology images tend to be less sensitive to the growth of amplitude of adversarial perturbations. (3) Percentage of successful attacks is growing with an increase of the number of iterations of the algorithm of generating adversarial perturbations with an asymptotic stabilization. (4) It was found that the success of attacks dropping dramatically when the original confidence of predicting image class exceeds 0.95. (5) The expected dependence of the percentage of successful attacks on the size of image training set was not confirmed.
http://arxiv.org/abs/1904.06964
We present a real-time stereo visual-inertial-SLAM system which is able to recover from complicatedkidnap scenarios and failures online in realtime. We propose to learn the whole-image-descriptorin a weakly supervised manner based on NetVLAD and decoupled convolutions. We analyse thetraining difficulties in using standard loss formulations and propose an allpairloss and show itseffect through extensive experiments. Compared to standard NetVLAD, our network takes an orderof magnitude fewer computations and model parameters, as a result runs about three times faster.We evaluate the representation power of our descriptor on standard datasets with precision-recall.Unlike previous loop detection methods which have been evaluated only on fronto-parallel revisits,we evaluate the performace of our method with competing methods on scenarios involving largeviewpoint difference. Finally, we present the fully functional system with relative computation andhandling of multiple world co-ordinate system which is able to reduce odometry drift, recover fromcomplicated kidnap scenarios and random odometry failures. We open source our fully functional system as an add-on for the popular VINS-Fusion.
http://arxiv.org/abs/1904.06962
Since it is difficult to collect face images of the same subject over a long range of age span, most existing face aging methods resort to unpaired datasets to learn age mappings. However, the matching ambiguity between young and aged face images inherent to unpaired training data may lead to unnatural changes of facial attributes during the aging process, which could not be solved by only enforcing identity consistency like most existing studies do. In this paper, we propose a attribute-aware face aging model with wavelet-based Generative Adversarial Networks (GANs) to address the above issues. To be specific, we embed facial attribute vectors into both generator and discriminator of the model to encourage each synthesized elderly face image to be faithful to the attribute of its corresponding input. In addition, a wavelet packet transform (WPT) module is incorporated to improve the visual fidelity of generated images by capturing age-related texture details at multiple scales in the frequency space. Qualitative results demonstrate the ability of our model to synthesize visually plausible face images, and extensive quantitative evaluation results show that the proposed method achieves state-of-the-art performance on existing datasets.
http://arxiv.org/abs/1809.06647
Recently, deep models have been successfully applied in several applications, especially with low-level representations. However, sparse, noisy samples and structured domains (with multiple objects and interactions) are some of the open challenges in most deep models. Column Networks, a deep architecture, can succinctly capture such domain structure and interactions, but may still be prone to sub-optimal learning from sparse and noisy samples. Inspired by the success of human-advice guided learning in AI, especially in data-scarce domains, we propose Knowledge-augmented Column Networks that leverage human advice/knowledge for better learning with noisy/sparse samples. Our experiments demonstrate that our approach leads to either superior overall performance or faster convergence (i.e., both effective and efficient).
http://arxiv.org/abs/1904.06950
In the Humanities and Social Sciences, there is increasing interest in approaches to information extraction, prediction, intelligent linkage, and dimension reduction applicable to large text corpora. With approaches in these fields being grounded in traditional statistical techniques, the need arises for frameworks whereby advanced NLP techniques such as topic modelling may be incorporated within classical methodologies. This paper provides a classical, supervised, statistical learning framework for prediction from text, using topic models as a data reduction method and the topics themselves as predictors, alongside typical statistical tools for predictive modelling. We apply this framework in a Social Sciences context (applied animal behaviour) as well as a Humanities context (narrative analysis) as examples of this framework. The results show that topic regression models perform comparably to their much less efficient equivalents that use individual words as predictors.
http://arxiv.org/abs/1904.06941
Autonomous navigation is an essential capability of smart mobility for mobile robots. Traditional methods must have the environment map to plan a collision-free path in workspace. Deep reinforcement learning (DRL) is a promising technique to realize the autonomous navigation task without a map, with which deep neural network can fit the mapping from observation to reasonable action through explorations. It should not only memorize the trained target, but more importantly, the planner can reason out the unseen goal. We proposed a new motion planner based on deep reinforcement learning that can arrive at new targets that have not been trained before in the indoor environment with RGB image and odometry only. The model has a structure of stacked Long Short-Term memory (LSTM). Finally, experiments were implemented in both simulated and real environments. The source code is available: https://github.com/marooncn/navbot.
http://arxiv.org/abs/1904.06933
Recent developed deep unsupervised methods allow us to jointly learn representation and cluster unlabelled data. These deep clustering methods %like DAC start with mainly focus on the correlation among samples, e.g., selecting high precision pairs to gradually tune the feature representation, which neglects other useful correlations. In this paper, we propose a novel clustering framework, named deep comprehensive correlation mining(DCCM), for exploring and taking full advantage of various kinds of correlations behind the unlabeled data from three aspects: 1) Instead of only using pair-wise information, pseudo-label supervision is proposed to investigate category information and learn discriminative features. 2) The features’ robustness to image transformation of input space is fully explored, which benefits the network learning and significantly improves the performance. 3) The triplet mutual information among features is presented for clustering problem to lift the recently discovered instance-level deep mutual information to a triplet-level formation, which further helps to learn more discriminative features. Extensive experiments on several challenging datasets show that our method achieves good performance, e.g., attaining $62.3\%$ clustering accuracy on CIFAR-10, and $34.0\%$ on CIFAR-100, both of which significantly surpass the state-of-the-art results more than $10.0\%$.
http://arxiv.org/abs/1904.06925
Crowd gatherings at social and cultural events are increasing in leaps and bounds with the increase in population. Surveillance through computer vision and expert decision making systems can help to understand the crowd phenomena at large gatherings. Understanding crowd phenomena can be helpful in early identification of unwanted incidents and their prevention. Motion flow is one of the important crowd phenomena that can be instrumental in describing the crowd behavior. Flows can be useful in understanding instabilities in the crowd. However, extracting motion flows is a challenging task due to randomness in crowd movement and limitations of the sensing device. Moreover, low-level features such as optical flow can be misleading if the randomness is high. In this paper, we propose a new model based on Langevin equation to analyze the linear dominant flows in videos of densely crowded scenarios. We assume a force model with three components, namely external force, confinement/drift force, and disturbance force. These forces are found to be sufficient to describe the linear or near-linear motion in dense crowd videos. The method is significantly faster as compared to existing popular crowd segmentation methods. The evaluation of the proposed model has been carried out on publicly available datasets as well as using our dataset. It has been observed that the proposed method is able to estimate and segment the linear flows in the dense crowd with better accuracy as compared to state-of-the-art techniques with substantial decrease in the computational overhead.
http://arxiv.org/abs/1904.07233
Long Short-Term Memory (LSTM) networks have recently shown remarkable performance in several tasks dealing with natural language generation, such as image captioning or poetry composition. Yet, only few works have analyzed text generated by LSTMs in order to quantitatively evaluate to which extent such artificial texts resemble those generated by humans. We compared the statistical structure of LSTM-generated language to that of written natural language, and to those produced by Markov models of various orders. In particular, we characterized the statistical structure of language by assessing word-frequency statistics, long-range correlations, and entropy measures. Our main finding is that while both LSTM and Markov-generated texts can exhibit features similar to real ones in their word-frequency statistics and entropy measures, LSTM-texts are shown to reproduce long-range correlations at scales comparable to those found in natural language. Moreover, for LSTM networks a temperature-like parameter controlling the generation process shows an optimal value—for which the produced texts are closest to real language—consistent across all the different statistical features investigated.
https://arxiv.org/abs/1804.04087
In image-to-image translation the goal is to learn a mapping from one image domain to another. Supervised approaches learn the mapping from paired samples. However, collecting large sets of image pairs is often prohibitively expensive or infeasible. In our work, we show that even training on the pairs implicitly, boosts the performance of unsupervised techniques by over 14% across several measurements. We illustrate that the injection of implicit pairs into unpaired sets strengthens the mapping between the two domains and improves the compatibility of their distributions. Furthermore, we show that for this purpose the implicit pairs can be pseudo-pairs, i.e., paired samples which only approximate a real pair. We demonstrate the effect of the approximated implicit samples on image-to-image translation problems, where such pseudo-pairs can be synthesized in one direction, but not in the other. We further show that pseudo-pairs are significantly more effective as implicit pairs in an unpaired setting, than directly using them explicitly in a paired setting.
http://arxiv.org/abs/1904.06913
We propose a novel regularization algorithm to train deep neural networks, in which data at training time is severely biased. Since a neural network efficiently learns data distribution, a network is likely to learn the bias information to categorize input data. It leads to poor performance at test time, if the bias is, in fact, irrelevant to the categorization. In this paper, we formulate a regularization loss based on mutual information between feature embedding and bias. Based on the idea of minimizing this mutual information, we propose an iterative algorithm to unlearn the bias information. We employ an additional network to predict the bias distribution and train the network adversarially against the feature embedding network. At the end of learning, the bias prediction network is not able to predict the bias not because it is poorly trained, but because the feature embedding network successfully unlearns the bias information. We also demonstrate quantitative and qualitative experimental results which show that our algorithm effectively removes the bias information from feature embedding.
http://arxiv.org/abs/1812.10352
State-of-the-art methods for counting people in crowded scenes rely on deep networks to estimate crowd density. They typically use the same filters over the whole image or over large image patches. Only then do they estimate local scale to compensate for perspective distortion. This is typically achieved by training an auxiliary classifier to select, for predefined image patches, the best kernel size among a limited set of choices. As such, these methods are not end-to-end trainable and restricted in the scope of context they can leverage. In this paper, we introduce an end-to-end trainable deep architecture that combines features obtained using multiple receptive field sizes and learns the importance of each such feature at each image location. In other words, our approach adaptively encodes the scale of the contextual information required to accurately predict crowd density. This yields an algorithm that outperforms state-of-the-art crowd counting methods, especially when perspective effects are strong.
http://arxiv.org/abs/1811.10452
Most of the classical denoising methods restore clear results by selecting and averaging pixels in the noisy input. Instead of relying on hand-crafted selecting and averaging strategies, we propose to explicitly learn this process with deep neural networks. Specifically, we propose deformable 2D kernels for image denoising where the sampling locations and kernel weights are both learned. The proposed kernel naturally adapts to image structures and could effectively reduce the oversmoothing artifacts. Furthermore, we develop 3D deformable kernels for video denoising to more efficiently sample pixels across the spatial-temporal space. Our method is able to solve the misalignment issues of large motion from dynamic scenes. For better training our video denoising model, we introduce the trilinear sampler and a new regularization term. We demonstrate that the proposed method performs favorably against the state-of-the-art image and video denoising approaches on both synthetic and real-world data.
http://arxiv.org/abs/1904.06903
In this paper, we present a novel guidance scheme based on model-based deep reinforcement learning (RL) technique. With model-based deep RL method, a deep neural network is trained as a predictive model of guidance dynamics which is incorporated into a model predictive path integral (MPPI) control framework. However the traditional MPPI framework assumes the actual environment similar to the training dataset for the deep neural network which is impractical in practice with different maneuvering of target, other perturbations and actuator failures. To address this problem, our method utilize meta-learning technique to make the deep neural dynamics model adapt to such changes online. With this approach we can alleviate the performance deterioration of standard MPPI control caused by the difference between actual environment and training data. Then, a novel guidance law for a varying velocity interceptor intercepting maneuvering target with desired terminal impact angle under actuator failure is constructed based on aforementioned techniques. Simulation and experiment results under different cases show the effectiveness and robustness of the proposed guidance law in achieving successful interceptions of maneuvering target.
http://arxiv.org/abs/1904.06892
The presented algorithms for segmentation and tracking follow a 3-step approach where we detect, track and finally segment nuclei. In the preprocessing phase, we detect centroids of the cell nuclei using a convolutional neural network (CNN) for the 2D images and a Laplacian-of-Gaussian Scale Space Maximum Projection approach for the 3D data sets. Tracking was performed in a backwards fashion on the predicted seed points, i.e., starting at the last frame and sequentially connecting corresponding objects until the first frame was reached. Correspondences were identified by propagating detections of a frame t to its preceding frame t-1 and by combining redundant detections using a hierarchical clustering approach. The tracked centroids were then used as input to variants of the seeded watershed algorithm to obtain the final segmentation.
http://arxiv.org/abs/1904.06890
We seek to detect visual relations in images of the form of triplets t = (subject, predicate, object), such as “person riding dog”, where training examples of the individual entities are available but their combinations are unseen at training. This is an important set-up due to the combinatorial nature of visual relations : collecting sufficient training data for all possible triplets would be very hard. The contributions of this work are three-fold. First, we learn a representation of visual relations that combines (i) individual embeddings for subject, object and predicate together with (ii) a visual phrase embedding that represents the relation triplet. Second, we learn how to transfer visual phrase embeddings from existing training triplets to unseen test triplets using analogies between relations that involve similar objects. Third, we demonstrate the benefits of our approach on three challenging datasets : on HICO-DET, our model achieves significant improvement over a strong baseline over both frequent and unseen triplets, and we confirm this improvement on the retrieval of unseen triplets with out-of-vocabulary predicates on COCO-a, as well as on the challenging unusual triplets of UnRel dataset.
http://arxiv.org/abs/1812.05736
Traditional neural objection detection methods use multi-scale features that allow multiple detectors to perform detecting tasks independently and in parallel. At the same time, with the handling of the prior box, the algorithm’s ability to deal with scale invariance is enhanced. However, too many prior boxes and independent detectors will increase the computational redundancy of the detection algorithm. In this study, we introduce Dubox, a new one-stage approach that detects the objects without prior box. Working with multi-scale features, the designed dual scale residual unit makes dual scale detectors no longer run independently. The second scale detector learns the residual of the first. Dubox has enhanced the capacity of heuristic-guided that can further enable the first scale detector to maximize the detection of small targets and the second to detect objects that cannot be identified by the first one. Besides, for each scale detector, with the new classification-regression progressive strapped loss makes our process not based on prior boxes. Integrating these strategies, our detection algorithm has achieved excellent performance in terms of speed and accuracy. Extensive experiments on the VOC, COCO object detection benchmark have confirmed the effectiveness of this algorithm.
http://arxiv.org/abs/1904.06883
This paper addresses the problem of determining dense pixel correspondences between two images and its application to geometric correspondence verification in image retrieval. The main contribution is a geometric correspondence verification approach for re-ranking a shortlist of retrieved database images based on their dense pair-wise matching with the query image at a pixel level. We determine a set of cyclically consistent dense pixel matches between the pair of images and evaluate local similarity of matched pixels using neural network based image descriptors. Final re-ranking is based on a novel similarity function, which fuses the local similarity metric with a global similarity metric and a geometric consistency measure computed for the matched pixels. For dense matching our approach utilizes a modified version of a recently proposed dense geometric correspondence network (DGC-Net), which we also improve by optimizing the architecture. The proposed model and similarity metric compare favourably to the state-of-the-art image retrieval methods. In addition, we apply our method to the problem of long-term visual localization demonstrating promising results and generalization across datasets.
http://arxiv.org/abs/1904.06882
Interactive reinforcement learning has become an important apprenticeship approach to speed up convergence in classic reinforcement learning problems. In this regard, a variant of interactive reinforcement learning is policy shaping which uses a parent-like trainer to propose the next action to be performed and by doing so reduces the search space by advice. On some occasions, the trainer may be another artificial agent which in turn was trained using reinforcement learning methods to afterward becoming an advisor for other learner-agents. In this work, we analyze internal representations and characteristics of artificial agents to determine which agent may outperform others to become a better trainer-agent. Using a polymath agent, as compared to a specialist agent, an advisor leads to a larger reward and faster convergence of the reward signal and also to a more stable behavior in terms of the state visit frequency of the learner-agents. Moreover, we analyze system interaction parameters in order to determine how influential they are in the apprenticeship process, where the consistency of feedback is much more relevant when dealing with different learner obedience parameters.
http://arxiv.org/abs/1904.06879
The present paper describes a singing voice synthesis based on convolutional neural networks (CNNs). Singing voice synthesis systems based on deep neural networks (DNNs) are currently being proposed and are improving the naturalness of synthesized singing voices. In these systems, the relationship between musical score feature sequences and acoustic feature sequences extracted from singing voices is modeled by DNNs. Then, an acoustic feature sequence of an arbitrary musical score is output in units of frames by the trained DNNs, and a natural trajectory of a singing voice is obtained by using a parameter generation algorithm. As singing voices contain rich expression, a powerful technique to model them accurately is required. In the proposed technique, long-term dependencies of singing voices are modeled by CNNs. An acoustic feature sequence is generated in units of segments that consist of long-term frames, and a natural trajectory is obtained without the parameter generation algorithm. Experimental results in a subjective listening test show that the proposed architecture can synthesize natural sounding singing voices.
http://arxiv.org/abs/1904.06868
Behavioral decision theories aim to explain human behavior. Can they help predict it? An open tournament for prediction of human choices in fundamental economic decision tasks is presented. The results suggest that integration of certain behavioral theories as features in machine learning systems provides the best predictions. Surprisingly, the most useful theories for prediction build on basic properties of human and animal learning and are very different from mainstream decision theories that focus on deviations from rational choice. Moreover, we find that theoretical features should be based not only on qualitative behavioral insights (e.g. loss aversion), but also on quantitative behavioral foresights generated by functional descriptive models (e.g. Prospect Theory). Our analysis prescribes a recipe for derivation of explainable, useful predictions of human decisions.
http://arxiv.org/abs/1904.06866
To achieve a dexterous robotic manipulation, we need to endow our robot with tactile feedback capability, i.e. the ability to drive action based on tactile sensing. In this paper, we specifically address the challenge of tactile servoing, i.e. given the current tactile sensing and a target/goal tactile sensing –memorized from a successful task execution in the past– what is the action that will bring the current tactile sensing to move closer towards the target tactile sensing at the next time step. We develop a data-driven approach to acquire a dynamics model for tactile servoing by learning from demonstration. Moreover, our method represents the tactile sensing information as to lie on a surface –or a 2D manifold– and perform a manifold learning, making it applicable to any tactile skin geometry. We evaluate our method on a contact point tracking task using a robot equipped with a tactile finger. A video demonstrating our approach can be seen in https://youtu.be/0QK0-Vx7WkI
http://arxiv.org/abs/1811.03704
Existing methods for image captioning are usually trained by cross entropy loss, which leads to exposure bias and the inconsistency between the optimizing function and evaluation metrics. Recently it has been shown that these two issues can be addressed by incorporating techniques from reinforcement learning, where one of the popular techniques is the advantage actor-critic algorithm that calculates per-token advantage by estimating state value with a parametrized estimator at the cost of introducing estimation bias. In this paper, we estimate state value without using a parametrized value estimator. With the properties of image captioning, namely, the deterministic state transition function and the sparse reward, state value is equivalent to its preceding state-action value, and we reformulate advantage function by simply replacing the former with the latter. Moreover, the reformulated advantage is extended to n-step, which can generally increase the absolute value of the mean of reformulated advantage while lowering variance. Then two kinds of rollout are adopted to estimate state-action value, which we call self-critical n-step training. Empirically we find that our method can obtain better performance compared to the state-of-the-art methods that use the sequence level advantage and parametrized estimator respectively on the widely used MSCOCO benchmark.
http://arxiv.org/abs/1904.06861
Thermal images are mainly used to detect the presence of people at night or in bad lighting conditions, but perform poorly at daytime. To solve this problem, most state-of-the-art techniques employ a fusion network that uses features from paired thermal and color images. Instead, we propose to augment thermal images with their saliency maps, to serve as an attention mechanism for the pedestrian detector especially during daytime. We investigate how such an approach results in improved performance for pedestrian detection using only thermal images, eliminating the need for paired color images. For our experiments, we train the Faster R-CNN for pedestrian detection and report the added effect of saliency maps generated using static and deep methods (PiCA-Net and R3-Net). Our best performing model results in an absolute reduction of miss rate by 13.4% and 19.4% over the baseline in day and night images respectively. We also annotate and release pixel level masks of pedestrians on a subset of the KAIST Multispectral Pedestrian Detection dataset, which is a first publicly available dataset for salient pedestrian detection.
http://arxiv.org/abs/1904.06859
With the prevalence of accessible depth sensors, dynamic human body skeletons have attracted much attention as a robust modality for action recognition. Previous methods model skeletons based on RNN or CNN, which has limited expressive power for irregular skeleton joints. While graph convolutional networks (GCN) have been proposed to address irregular graph-structured data, the fundamental graph construction remains challenging. In this paper, we represent skeletons naturally on graphs, and propose a graph regression based GCN (GR-GCN) for skeleton-based action recognition, aiming to capture the spatio-temporal variation in the data. As the graph representation is crucial to graph convolution, we first propose graph regression to statistically learn the underlying graph from multiple observations. In particular, we provide spatio-temporal modeling of skeletons and pose an optimization problem on the graph structure over consecutive frames, which enforces the sparsity of the underlying graph for efficient representation. The optimized graph not only connects each joint to its neighboring joints in the same frame strongly or weakly, but also links with relevant joints in the previous and subsequent frames. We then feed the optimized graph into the GCN along with the coordinates of the skeleton sequence for feature learning, where we deploy high-order and fast Chebyshev approximation of spectral graph convolution. Further, we provide analysis of the variation characterization by the Chebyshev approximation. Experimental results validate the effectiveness of the proposed graph regression and show that the proposed GR-GCN achieves the state-of-the-art performance on the widely used NTU RGB+D, UT-Kinect and SYSU 3D datasets.
http://arxiv.org/abs/1811.12013
Sound frisson is a subjective experience wherein people tend to perceive the feeling of chills in addition to a physiological response, such as goosebumps. Multiple examples of frisson inducing sounds have been reported in the large online community, but the mechanism of sound frisson is still elusive. Typical frisson inducing sounds contain a looming effect, in which a sound seems to be approaching close to one’s peripersonal space. Previous studies on sound in peripersonal space have reported objective measurements of sound-inducing effects, but few studies have investigated the subjective experience of frisson-inducing sound. Here, we investigate whether sound stimulus moving around the human head can also produce subjective ratings of frisson. Our results show that the participants experienced sound-induced frisson when auditory stimuli were rotated around the head, regardless of the sound sources.
http://arxiv.org/abs/1904.06851
In this paper, we propose a novel matching based tracker by investigating the relationship between template matching and the recent popular correlation filter based trackers (CFTs). Compared to the correlation operation in CFTs, a sophisticated similarity metric termed “mutual buddies similarity” (MBS) is proposed to exploit the relationship of multiple reciprocal nearest neighbors for target matching. By doing so, our tracker obtains powerful discriminative ability on distinguishing target and background as demonstrated by both empirical and theoretical analyses. Besides, instead of utilizing single template with the improper updating scheme in CFTs, we design a novel online template updating strategy named “memory filtering” (MF), which aims to select a certain amount of representative and reliable tracking results in history to construct the current stable and expressive template set. This scheme is beneficial for the proposed tracker to comprehensively “understand” the target appearance variations, “recall” some stable results. Both qualitative and quantitative evaluations on two benchmarks suggest that the proposed tracking method performs favorably against some recently developed CFTs and other competitive trackers.
http://arxiv.org/abs/1904.06842
Particle Imaging Velocimetry (PIV) estimates the flow of fluid by analyzing the motion of injected particles. The problem is challenging as the particles lie at different depths but have similar appearance and tracking a large number of particles is particularly difficult. In this paper, we present a PIV solution that uses densely sampled light field to reconstruct and track 3D particles. We exploit the refocusing capability and focal symmetry constraint of the light field for reliable particle depth estimation. We further propose a new motion-constrained optical flow estimation scheme by enforcing local motion rigidity and the Navier-Stoke constraint. Comprehensive experiments on synthetic and real experiments show that using a single light field camera, our technique can recover dense and accurate 3D fluid flows in small to medium volumes.
http://arxiv.org/abs/1904.06841
Compared with global average pooling in existing deep convolutional neural networks (CNNs), global covariance pooling can capture richer statistics of deep features, having potential for improving representation and generalization abilities of deep CNNs. However, integration of global covariance pooling into deep CNNs brings two challenges: (1) robust covariance estimation given deep features of high dimension and small sample; (2) appropriate use of geometry of covariances. To address these challenges, we propose a global Matrix Power Normalized COVariance (MPN-COV) Pooling. Our MPN-COV conforms to a robust covariance estimator, very suitable for scenario of high dimension and small sample. It can also be regarded as power-Euclidean metric between covariances, effectively exploiting their geometry. Furthermore, a global Gaussian embedding method is proposed to incorporate first-order statistics into MPN-COV. For fast training of MPN-COV networks, we propose an iterative matrix square root normalization, avoiding GPU unfriendly eigen-decomposition inherent in MPN-COV. Additionally, progressive 1x1 and group convolutions are introduced to compact covariance representations. The MPN-COV and its variants are highly modular, readily plugged into existing deep CNNs. Extensive experiments are conducted on large-scale object classification, scene categorization, fine-grained visual recognition and texture classification, showing our methods are superior to the counterparts and achieve state-of-the-art performance.
http://arxiv.org/abs/1904.06836
Globally normalized neural sequence models are considered superior to their locally normalized equivalents because they may ameliorate the effects of label bias. However, when considering high-capacity neural parametrizations that condition on the whole input sequence, both model classes are theoretically equivalent in terms of the distributions they are capable of representing. Thus, the practical advantage of global normalization in the context of modern neural methods remains unclear. In this paper, we attempt to shed light on this problem through an empirical study. We extend an approach for search-aware training via a continuous relaxation of beam search (Goyal et al., 2017b) in order to enable training of globally normalized recurrent sequence models through simple backpropagation. We then use this technique to conduct an empirical study of the interaction between global normalization, high-capacity encoders, and search-aware optimization. We observe that in the context of inexact search, globally normalized neural models are still more effective than their locally normalized counterparts. Further, since our training approach is sensitive to warm-starting with pre-trained models, we also propose a novel initialization strategy based on self-normalization for pre-training globally normalized models. We perform analysis of our approach on two tasks: CCG supertagging and Machine Translation, and demonstrate the importance of global normalization under different conditions while using search-aware training.
http://arxiv.org/abs/1904.06834
Grasping and manipulating objects is an important human skill. Since hand-object contact is fundamental to grasping, capturing it can lead to important insights. However, observing contact through external sensors is challenging because of occlusion and the complexity of the human hand. We present ContactDB, a novel dataset of contact maps for household objects that captures the rich hand-object contact that occurs during grasping, enabled by use of a thermal camera. Participants in our study grasped 3D printed objects with a post-grasp functional intent. ContactDB includes 3750 3D meshes of 50 household objects textured with contact maps and 375K frames of synchronized RGB-D+thermal images. To the best of our knowledge, this is the first large-scale dataset that records detailed contact maps for human grasps. Analysis of this data shows the influence of functional intent and object size on grasping, the tendency to touch/avoid ‘active areas’, and the high frequency of palm and proximal finger contact. Finally, we train state-of-the-art image translation and 3D convolution algorithms to predict diverse contact patterns from object shape. Data, code and models are available at https://contactdb.cc.gatech.edu.
http://arxiv.org/abs/1904.06830
We tackle the problem of generating a pun sentence given a pair of homophones (e.g., “died” and “dyed”). Supervised text generation is inappropriate due to the lack of a large corpus of puns, and even if such a corpus existed, mimicry is at odds with generating novel content. In this paper, we propose an unsupervised approach to pun generation using a corpus of unhumorous text and what we call the local-global surprisal principle: we posit that in a pun sentence, there is a strong association between the pun word (e.g., “dyed”) and the distant context, as well as a strong association between the alternative word (e.g., “died”) and the immediate context. This contrast creates surprise and thus humor. We instantiate this principle for pun generation in two ways: (i) as a measure based on the ratio of probabilities under a language model, and (ii) a retrieve-and-edit approach based on words suggested by a skip-gram model. Human evaluation shows that our retrieve-and-edit approach generates puns successfully 31% of the time, tripling the success rate of a neural generation baseline.
http://arxiv.org/abs/1904.06828