Photo composition is an important factor affecting the aesthetics in photography. However, it is a highly challenging task to model the aesthetic properties of good compositions due to the lack of globally applicable rules to the wide variety of photographic styles. Inspired by the thinking process of photo taking, we formulate the photo composition problem as a view finding process which successively examines pairs of views and determines their aesthetic preferences. We further exploit the rich professional photographs on the web to mine unlimited high-quality ranking samples and demonstrate that an aesthetics-aware deep ranking network can be trained without explicitly modeling any photographic rules. The resulting model is simple and effective in terms of its architectural design and data sampling method. It is also generic since it naturally learns any photographic rules implicitly encoded in professional photographs. The experiments show that the proposed view finding network achieves state-of-the-art performance with sliding window search strategy on two image cropping datasets.
https://arxiv.org/abs/1702.00503
Most neural machine translation (NMT) models are based on the sequential encoder-decoder framework, which makes no use of syntactic information. In this paper, we improve this model by explicitly incorporating source-side syntactic trees. More specifically, we propose (1) a bidirectional tree encoder which learns both sequential and tree structured representations; (2) a tree-coverage model that lets the attention depend on the source-side syntax. Experiments on Chinese-English translation demonstrate that our proposed models outperform the sequential attentional model as well as a stronger baseline with a bottom-up tree encoder and word coverage.
https://arxiv.org/abs/1707.05436
We investigate the utility of different auxiliary objectives and training strategies within a neural sequence labeling approach to error detection in learner writing. Auxiliary costs provide the model with additional linguistic information, allowing it to learn general-purpose compositional features that can then be exploited for other objectives. Our experiments show that a joint learning approach trained with parallel labels on in-domain data improves performance over the previous best error detection system. While the resulting model has the same number of parameters, the additional objectives allow it to be optimised more efficiently and achieve better performance.
https://arxiv.org/abs/1707.05227
We present an analytical framework that allows the quantitative study of statistical dynamic properties of networks with adaptive nodes that have memory and is used to examine the emergence of oscillations in networks with response failures. The frequency of the oscillations was quantitatively found to increase with the excitability of the nodes and with the average degree of the network and to decrease with delays between nodes. For networks of networks, diverse cluster oscillation modes were found as a function of the topology. Analytical results are in agreement with large-scale simulations and open the horizon for understanding network dynamics composed of finite memory nodes as well as their different phases of activity.
https://arxiv.org/abs/1707.05157
Since convolutional neural network(CNN)models emerged,several tasks in computer vision have actively deployed CNN models for feature extraction. However,the conventional CNN models have a high computational cost and require high memory capacity, which is impractical and unaffordable for commercial applications such as real-time on-road object detection on embedded boards or mobile platforms. To tackle this limitation of CNN models, this paper proposes a wide-residual-inception (WR-Inception) network, which constructs the architecture based on a residual inception unit that captures objects of various sizes on the same feature map, as well as shallower and wider layers, compared to state-of-the-art networks like ResNet. To verify the proposed networks, this paper conducted two experiments; one is a classification task on CIFAR-10/100 and the other is an on-road object detection task using a Single-Shot Multi-box Detector(SSD) on the KITTI dataset.
https://arxiv.org/abs/1702.01243
Recent years have witnessed a resurgence of interest in video summarization. However, one of the main obstacles to the research on video summarization is the user subjectivity - users have various preferences over the summaries. The subjectiveness causes at least two problems. First, no single video summarizer fits all users unless it interacts with and adapts to the individual users. Second, it is very challenging to evaluate the performance of a video summarizer. To tackle the first problem, we explore the recently proposed query-focused video summarization which introduces user preferences in the form of text queries about the video into the summarization process. We propose a memory network parameterized sequential determinantal point process in order to attend the user query onto different video frames and shots. To address the second challenge, we contend that a good evaluation metric for video summarization should focus on the semantic information that humans can perceive rather than the visual features or temporal overlaps. To this end, we collect dense per-video-shot concept annotations, compile a new dataset, and suggest an efficient evaluation method defined upon the concept annotations. We conduct extensive experiments contrasting our video summarizer to existing ones and present detailed analyses about the dataset and the new evaluation method.
https://arxiv.org/abs/1707.04960
This work presents a novel approach to Automatic Post-Editing (APE) and Word-Level Quality Estimation (QE) using ensembles of specialized Neural Machine Translation (NMT) systems. Word-level features that have proven effective for QE are included as input factors, expanding the representation of the original source and the machine translation hypothesis, which are used to generate an automatically post-edited hypothesis. We train a suite of NMT models that use different input representations, but share the same output space. These models are then ensembled together, and tuned for both the APE and the QE task. We thus attempt to connect the state-of-the-art approaches to APE and QE within a single framework. Our models achieve state-of-the-art results in both tasks, with the only difference in the tuning step which learns weights for each component of the ensemble.
https://arxiv.org/abs/1706.05083
This is part III of three-part work. In parts I and II, we have presented eight variants for simplified Long Short Term Memory (LSTM) recurrent neural networks (RNNs). It is noted that fast computation, specially in constrained computing resources, are an important factor in processing big time-sequence data. In this part III paper, we present and evaluate two new LSTM model variants which dramatically reduce the computational load while retaining comparable performance to the base (standard) LSTM RNNs. In these new variants, we impose (Hadamard) pointwise state multiplications in the cell-memory network in addition to the gating signal networks.
https://arxiv.org/abs/1707.04626
This is part II of three-part work. Here, we present a second set of inter-related five variants of simplified Long Short-term Memory (LSTM) recurrent neural networks by further reducing adaptive parameters. Two of these models have been introduced in part I of this work. We evaluate and verify our model variants on the benchmark MNIST dataset and assert that these models are comparable to the base LSTM model while use progressively less number of parameters. Moreover, we observe that in case of using the ReLU activation, the test accuracy performance of the standard LSTM will drop after a number of epochs when learning parameter become larger. However all of the new model variants sustain their performance.
https://arxiv.org/abs/1707.04623
We present five variants of the standard Long Short-term Memory (LSTM) recurrent neural networks by uniformly reducing blocks of adaptive parameters in the gating mechanisms. For simplicity, we refer to these models as LSTM1, LSTM2, LSTM3, LSTM4, and LSTM5, respectively. Such parameter-reduced variants enable speeding up data training computations and would be more suitable for implementations onto constrained embedded platforms. We comparatively evaluate and verify our five variant models on the classical MNIST dataset and demonstrate that these variant models are comparable to a standard implementation of the LSTM model while using less number of parameters. Moreover, we observe that in some cases the standard LSTM’s accuracy performance will drop after a number of epochs when using the ReLU nonlinearity; in contrast, however, LSTM3, LSTM4 and LSTM5 will retain their performance.
https://arxiv.org/abs/1707.04619
In this paper, we describe our submissions to the WMT17 Multimodal Translation Task. For Task 1 (multimodal translation), our best scoring system is a purely textual neural translation of the source image caption to the target language. The main feature of the system is the use of additional data that was acquired by selecting similar sentences from parallel corpora and by data synthesis with back-translation. For Task 2 (cross-lingual image captioning), our best submitted system generates an English caption which is then translated by the best system used in Task 1. We also present negative results, which are based on ideas that we believe have potential of making improvements, but did not prove to be useful in our particular setup.
https://arxiv.org/abs/1707.04550
Urban informatics explore data science methods to address different urban issues intensively based on data. The large variety and quantity of data available should be explored but this brings important challenges. For instance, although there are powerful computer vision methods that may be explored, they may require large annotated datasets. In this work we propose a novel approach to automatically creating an object recognition system with minimal manual annotation. The basic idea behind the method is to use large input datasets using available online cameras on large cities. A off-the-shelf weak classifier is used to detect an initial set of urban elements of interest (e.g. cars, pedestrians, bikes, etc.). Such initial dataset undergoes a quality control procedure and it is subsequently used to fine tune a strong classifier. Quality control and comparative performance assessment are used as part of the pipeline. We evaluate the method for detecting cars based on monitoring cameras. Experimental results using real data show that despite losing generality, the final detector provides better detection rates tailored to the selected cameras. The programmed robot gathered 770 video hours from 24 online city cameras (~300GB), which has been fed to the proposed system. Our approach has shown that the method nearly doubled the recall (93\%) with respect to state-of-the-art methods using off-the-shelf algorithms.
https://arxiv.org/abs/1706.09308
Using image context is an effective approach for improving object detection. Previously proposed methods used contextual cues that rely on semantic or spatial information. In this work, we explore a different kind of contextual information: inner-scene similarity. We present the CISS (Context by Inner Scene Similarity) algorithm, which is based on the observation that two visually similar sub-image patches are likely to share semantic identities, especially when both appear in the same image. CISS uses base-scores provided by a base detector and performs as a post-detection stage. For each candidate sub-image (denoted anchor), the CISS algorithm finds a few similar sub-images (denoted supporters), and, using them, calculates a new enhanced score for the anchor. This is done by utilizing the base-scores of the supporters and a pre-trained dependency model. The new scores are modeled as a linear function of the base scores of the anchor and the supporters and is estimated using a minimum mean square error optimization. This approach results in: (a) improved detection of partly occluded objects (when there are similar non-occluded objects in the scene), and (b) fewer false alarms (when the base detector mistakenly classifies a background patch as an object). This work relates to Duncan and Humphreys’ “similarity theory,” a psychophysical study. which suggested that the human visual system perceptually groups similar image regions and that the classification of one region is affected by the estimated identity of the other. Experimental results demonstrate the enhancement of a base detector’s scores on the PASCAL VOC dataset.
https://arxiv.org/abs/1707.04406
While convolutional neural networks have gained impressive success recently in solving structured prediction problems such as semantic segmentation, it remains a challenge to differentiate individual object instances in the scene. Instance segmentation is very important in a variety of applications, such as autonomous driving, image captioning, and visual question answering. Techniques that combine large graphical models with low-level vision have been proposed to address this problem; however, we propose an end-to-end recurrent neural network (RNN) architecture with an attention mechanism to model a human-like counting process, and produce detailed instance segmentations. The network is jointly trained to sequentially produce regions of interest as well as a dominant object segmentation within each region. The proposed model achieves competitive results on the CVPPP, KITTI, and Cityscapes datasets.
https://arxiv.org/abs/1605.09410
Hybrid MKNF knowledge bases have been considered one of the dominant approaches to combining open world ontology languages with closed world rule-based languages. Currently, the only known inference methods are based on the approach of guess-and-verify, while most modern SAT/ASP solvers are built under the DPLL architecture. The central impediment here is that it is not clear what constitutes a constraint propagator, a key component employed in any DPLL-based solver. In this paper, we address this problem by formulating the notion of unfounded sets for nondisjunctive hybrid MKNF knowledge bases, based on which we propose and study two new well-founded operators. We show that by employing a well-founded operator as a constraint propagator, a sound and complete DPLL search engine can be readily defined. We compare our approach with the operator based on the alternating fixpoint construction by Knorr et al [2011] and show that, when applied to arbitrary partial partitions, the new well-founded operators not only propagate more truth values but also circumvent the non-converging behavior of the latter. In addition, we study the possibility of simplifying a given hybrid MKNF knowledge base by employing a well-founded operator, and show that, out of the two operators proposed in this paper, the weaker one can be applied for this purpose and the stronger one cannot. These observations are useful in implementing a grounder for hybrid MKNF knowledge bases, which can be applied before the computation of MKNF models. The paper is under consideration for acceptance in TPLP.
https://arxiv.org/abs/1707.01959
We have calculated the key characteristics of associative (content-addressable) spatial-temporal memories based on neuromorphic networks with restricted connectivity - “CrossNets”. Such networks may be naturally implemented in nanoelectronic hardware using hybrid CMOS/memristor circuits, which may feature extremely high energy efficiency, approaching that of biological cortical circuits, at much higher operation speed. Our numerical simulations, in some cases confirmed by analytical calculations, have shown that the characteristics depend substantially on the method of information recording into the memory. Of the four methods we have explored, two look especially promising - one based on the quadratic programming, and the other one being a specific discrete version of the gradient descent. The latter method provides a slightly lower memory capacity (at the same fidelity) then the former one, but it allows local recording, which may be more readily implemented in nanoelectronic hardware. Most importantly, at the synchronous retrieval, both methods provide a capacity higher than that of the well-known Ternary Content-Addressable Memories with the same number of nonvolatile memory cells (e.g., memristors), though the input noise immunity of the CrossNet memories is somewhat lower.
https://arxiv.org/abs/1707.03855
It has been shown that most machine learning algorithms are susceptible to adversarial perturbations. Slightly perturbing an image in a carefully chosen direction in the image space may cause a trained neural network model to misclassify it. Recently, it was shown that physical adversarial examples exist: printing perturbed images then taking pictures of them would still result in misclassification. This raises security and safety concerns. However, these experiments ignore a crucial property of physical objects: the camera can view objects from different distances and at different angles. In this paper, we show experiments that suggest that current constructions of physical adversarial examples do not disrupt object detection from a moving platform. Instead, a trained neural network classifies most of the pictures taken from different distances and angles of a perturbed image correctly. We believe this is because the adversarial property of the perturbation is sensitive to the scale at which the perturbed picture is viewed, so (for example) an autonomous car will misclassify a stop sign only from a small range of distances. Our work raises an important question: can one construct examples that are adversarial for many or most viewing conditions? If so, the construction should offer very significant insights into the internal representation of patterns by deep networks. If not, there is a good prospect that adversarial examples can be reduced to a curiosity with little practical impact.
https://arxiv.org/abs/1707.03501
A model of bit-strings, that uses the technique of multi-spin coding, was previously used to study the time evolution of B-cell clone repertoire, in a paper by Lagreca, Almeida and Santos. In this work we extend that simplified model to include independently the role of the populations of antibodies, in the control of the immune response, producing mechanisms of differentiation and regulation in a more complete way. Although the antibodies have the same molecular shape of the B-cells receptors (BCR), they should present a different time evolution and thus should be treated separately. We have also studied a possible model for the network immune memory, suggesting a random memory regeneration, which is self-perpetuating.
https://arxiv.org/abs/1707.09295
Procedural terrain generation for video games has been traditionally been done with smartly designed but handcrafted algorithms that generate heightmaps. We propose a first step toward the learning and synthesis of these using recent advances in deep generative modelling with openly available satellite imagery from NASA.
https://arxiv.org/abs/1707.03383
Similarity search is a key to a variety of applications including content-based search for images and video, recommendation systems, data deduplication, natural language processing, computer vision, databases, computational biology, and computer graphics. At its core, similarity search manifests as k-nearest neighbors (kNN), a computationally simple primitive consisting of highly parallel distance calculations and a global top-k sort. However, kNN is poorly supported by today’s architectures because of its high memory bandwidth requirements. This paper proposes an application-driven near-data processing accelerator for similarity search: the Similarity Search Associative Memory (SSAM). By instantiating compute units close to memory, SSAM benefits from the higher memory bandwidth and density exposed by emerging memory technologies. We evaluate the SSAM design down to layout on top of the Micron hybrid memory cube (HMC), and show that SSAM can achieve up to two orders of magnitude area-normalized throughput and energy efficiency improvement over multicore CPUs; we also show SSAM is faster and more energy efficient than competing GPUs and FPGAs. Finally, we show that SSAM is also useful for other data intensive tasks like kNN index construction, and can be generalized to semantically function as a high capacity content addressable memory.
https://arxiv.org/abs/1606.03742
We discuss the properties of the dynamics of purely memristive circuits using a recently derived consistent equation for the internal memory variables of the involved memristors. In particular, we show that the number of independent memory states in a memristive circuit is constrained by the circuit conservation laws, and that the dynamics preserves these symmetries by means of a projection on the physical subspace. Moreover, we discuss other symmetries of the dynamics under various transformations of the internal memory, and study the linearized and strongly non-linear regimes of the dynamics. In the strongly non-linear regime, we derive a conservation law for the internal memory variables. We also provide a condition on the reality of the eigenvalues of Lyapunov matrices describing the linearized dynamics close to a fixed point. We show that the eigenvalues ca be imaginary only for mixtures of passive and active components. Our last result concerns the weak non-linear regime. We show that the internal memory dynamics can be interpreted as a constrained gradient descent, and provide the functional being minimized. This latter result provides another direct connection between memristors and learning.
https://arxiv.org/abs/1611.02104
We present a simple dynamic batching approach applicable to a large class of dynamic architectures that consistently yields speedups of over 10x. We provide performance bounds when the architecture is not known a priori and a stronger bound in the special case where the architecture is a predetermined balanced tree. We evaluate our approach on Johnson et al.’s recent visual question answering (VQA) result of his CLEVR dataset by Inferring and Executing Programs (IEP). We also evaluate on sparsely gated mixture of experts layers and achieve speedups of up to 1000x over the naive implementation.
https://arxiv.org/abs/1707.02402
In this paper, we propose a novel end-to-end approach for scalable visual search infrastructure. We discuss the challenges we faced for a massive volatile inventory like at eBay and present our solution to overcome those. We harness the availability of large image collection of eBay listings and state-of-the-art deep learning techniques to perform visual search at scale. Supervised approach for optimized search limited to top predicted categories and also for compact binary signature are key to scale up without compromising accuracy and precision. Both use a common deep neural network requiring only a single forward inference. The system architecture is presented with in-depth discussions of its basic components and optimizations for a trade-off between search relevance and latency. This solution is currently deployed in a distributed cloud infrastructure and fuels visual search in eBay ShopBot and Close5. We show benchmark on ImageNet dataset on which our approach is faster and more accurate than several unsupervised baselines. We share our learnings with the hope that visual search becomes a first class citizen for all large scale search engines rather than an afterthought.
https://arxiv.org/abs/1706.03154
Predicting the click-through rate of an advertisement is a critical component of online advertising platforms. In sponsored search, the click-through rate estimates the probability that a displayed advertisement is clicked by a user after she submits a query to the search engine. Commercial search engines typically rely on machine learning models trained with a large number of features to make such predictions. This is inevitably requires a lot of engineering efforts to define, compute, and select the appropriate features. In this paper, we propose two novel approaches (one working at character level and the other working at word level) that use deep convolutional neural networks to predict the click-through rate of a query-advertisement pair. Specially, the proposed architectures only consider the textual content appearing in a query-advertisement pair as input, and produce as output a click-through rate prediction. By comparing the character-level model with the word-level model, we show that language representation can be learnt from scratch at character level when trained on enough data. Through extensive experiments using billions of query-advertisement pairs of a popular commercial search engine, we demonstrate that both approaches significantly outperform a baseline model built on well-selected text features and a state-of-the-art word2vec-based approach. Finally, by combining the predictions of the deep models introduced in this study with the prediction of the model in production of the same commercial search engine, we significantly improve the accuracy and the calibration of the click-through rate prediction of the production system.
https://arxiv.org/abs/1707.02158
Question answering is an important and difficult task in the natural language processing domain, because many basic natural language processing tasks can be cast into a question answering task. Several deep neural network architectures have been developed recently, which employ memory and inference components to memorize and reason over text information, and generate answers to questions. However, a major drawback of many such models is that they are capable of only generating single-word answers. In addition, they require large amount of training data to generate accurate answers. In this paper, we introduce the Long-Term Memory Network (LTMN), which incorporates both an external memory module and a Long Short-Term Memory (LSTM) module to comprehend the input data and generate multi-word answers. The LTMN model can be trained end-to-end using back-propagation and requires minimal supervision. We test our model on two synthetic data sets (based on Facebook’s bAbI data set) and the real-world Stanford question answering data set, and show that it can achieve state-of-the-art performance.
https://arxiv.org/abs/1707.01961
Hadoop and Spark are widely used distributed processing frameworks for large-scale data processing in an efficient and fault-tolerant manner on private or public clouds. These big-data processing systems are extensively used by many industries, e.g., Google, Facebook, and Amazon, for solving a large class of problems, e.g., search, clustering, log analysis, different types of join operations, matrix multiplication, pattern matching, and social network analysis. However, all these popular systems have a major drawback in terms of locally distributed computations, which prevent them in implementing geographically distributed data processing. The increasing amount of geographically distributed massive data is pushing industries and academia to rethink the current big-data processing systems. The novel frameworks, which will be beyond state-of-the-art architectures and technologies involved in the current system, are expected to process geographically distributed data at their locations without moving entire raw datasets to a single location. In this paper, we investigate and discuss challenges and requirements in designing geographically distributed data processing frameworks and protocols. We classify and study batch processing (MapReduce-based systems), stream processing (Spark-based systems), and SQL-style processing geo-distributed frameworks, models, and algorithms with their overhead issues.
https://arxiv.org/abs/1707.01869
We present RON, an efficient and effective framework for generic object detection. Our motivation is to smartly associate the best of the region-based (e.g., Faster R-CNN) and region-free (e.g., SSD) methodologies. Under fully convolutional architecture, RON mainly focuses on two fundamental problems: (a) multi-scale object localization and (b) negative sample mining. To address (a), we design the reverse connection, which enables the network to detect objects on multi-levels of CNNs. To deal with (b), we propose the objectness prior to significantly reduce the searching space of objects. We optimize the reverse connection, objectness prior and object detector jointly by a multi-task loss function, thus RON can directly predict final detection results from all locations of various feature maps. Extensive experiments on the challenging PASCAL VOC 2007, PASCAL VOC 2012 and MS COCO benchmarks demonstrate the competitive performance of RON. Specifically, with VGG-16 and low resolution 384X384 input size, the network gets 81.3% mAP on PASCAL VOC 2007, 80.7% mAP on PASCAL VOC 2012 datasets. Its superiority increases when datasets become larger and more difficult, as demonstrated by the results on the MS COCO dataset. With 1.5G GPU memory at test phase, the speed of the network is 15 FPS, 3X faster than the Faster R-CNN counterpart.
https://arxiv.org/abs/1707.01691
Neural network acoustic models have significantly advanced state of the art speech recognition over the past few years. However, they are usually computationally expensive due to the large number of matrix-vector multiplications and nonlinearity operations. Neural network models also require significant amounts of memory for inference because of the large model size. For these two reasons, it is challenging to deploy neural network based speech recognizers on resource-constrained platforms such as embedded devices. This paper investigates the use of binary weights and activations for computation and memory efficient neural network acoustic models. Compared to real-valued weight matrices, binary weights require much fewer bits for storage, thereby cutting down the memory footprint. Furthermore, with binary weights or activations, the matrix-vector multiplications are turned into addition and subtraction operations, which are computationally much faster and more energy efficient for hardware platforms. In this paper, we study the applications of binary weights and activations for neural network acoustic modeling, reporting encouraging results on the WSJ and AMI corpora.
https://arxiv.org/abs/1706.09453
In state-of-the-art Neural Machine Translation (NMT), an attention mechanism is used during decoding to enhance the translation. At every step, the decoder uses this mechanism to focus on different parts of the source sentence to gather the most useful information before outputting its target word. Recently, the effectiveness of the attention mechanism has also been explored for multimodal tasks, where it becomes possible to focus both on sentence parts and image regions that they describe. In this paper, we compare several attention mechanism on the multimodal translation task (English, image to German) and evaluate the ability of the model to make use of images to improve translation. We surpass state-of-the-art scores on the Multi30k data set, we nevertheless identify and report different misbehavior of the machine while translating.
https://arxiv.org/abs/1707.00995
Question-answering (QA) on video contents is a significant challenge for achieving human-level intelligence as it involves both vision and language in real-world settings. Here we demonstrate the possibility of an AI agent performing video story QA by learning from a large amount of cartoon videos. We develop a video-story learning model, i.e. Deep Embedded Memory Networks (DEMN), to reconstruct stories from a joint scene-dialogue video stream using a latent embedding space of observed data. The video stories are stored in a long-term memory component. For a given question, an LSTM-based attention model uses the long-term memory to recall the best question-story-answer triplet by focusing on specific words containing key information. We trained the DEMN on a novel QA dataset of children’s cartoon video series, Pororo. The dataset contains 16,066 scene-dialogue pairs of 20.5-hour videos, 27,328 fine-grained sentences for scene description, and 8,913 story-related QA pairs. Our experimental results show that the DEMN outperforms other QA models. This is mainly due to 1) the reconstruction of video stories in a scene-dialogue combined form that utilize the latent embedding and 2) attention. DEMN also achieved state-of-the-art results on the MovieQA benchmark.
https://arxiv.org/abs/1707.00836
Despite recent advances, large scale visual artifacts are still a common occurrence in images generated by GANs. Previous work has focused on improving the generator’s capability to accurately imitate the data distribution $p_{data}$. In this paper, we instead explore methods that enable GANs to actively avoid errors by manipulating the input space. The core idea is to apply small changes to each noise vector in order to shift them away from areas in the input space that tend to result in errors. We derive three different architectures from that idea. The main one of these consists of a simple residual module that leads to significantly less visual artifacts, while only slightly decreasing diversity. The module is trivial to add to existing GANs and costs almost zero computation and memory.
https://arxiv.org/abs/1707.00768
A popular testbed for deep learning has been multimodal recognition of human activity or gesture involving diverse inputs such as video, audio, skeletal pose and depth images. Deep learning architectures have excelled on such problems due to their ability to combine modality representations at different levels of nonlinear feature extraction. However, designing an optimal architecture in which to fuse such learned representations has largely been a non-trivial human engineering effort. We treat fusion structure optimization as a hyper-parameter search and cast it as a discrete optimization problem under the Bayesian optimization framework. We propose a novel graph-induced kernel to compute structural similarities in the search space of tree-structured multimodal architectures and demonstrate its effectiveness using two challenging multimodal human activity recognition datasets.
https://arxiv.org/abs/1707.00750
In this paper, we propose a new approach for retrieval of video segments using natural language queries. Unlike most previous approaches such as concept-based methods or rule-based structured models, the proposed method uses image captioning model to construct sentential queries for visual information. In detail, our approach exploits multiple captions generated by visual features in each image with `Densecap’. Then, the similarities between captions of adjacent images are calculated, which is used to track semantically similar captions over multiple frames. Besides introducing this novel idea of ‘tracking by captioning’, the proposed method is one of the first approaches that uses a language generation model learned by neural networks to construct semantic query describing the relations and properties of visual information. To evaluate the effectiveness of our approach, we have created a new evaluation dataset, which contains about 348 segments of scenes in 20 movie-trailers. Through quantitative and qualitative evaluation, we show that our method is effective for retrieval of video segments using natural language queries.
https://arxiv.org/abs/1707.00251
Learning a classifier from ambiguously labeled face images is challenging since training images are not always explicitly-labeled. For instance, face images of two persons in a news photo are not explicitly labeled by their names in the caption. We propose a Matrix Completion for Ambiguity Resolution (MCar) method for predicting the actual labels from ambiguously labeled images. This step is followed by learning a standard supervised classifier from the disambiguated labels to classify new images. To prevent the majority labels from dominating the result of MCar, we generalize MCar to a weighted MCar (WMCar) that handles label imbalance. Since WMCar outputs a soft labeling vector of reduced ambiguity for each instance, we can iteratively refine it by feeding it as the input to WMCar. Nevertheless, such an iterative implementation can be affected by the noisy soft labeling vectors, and thus the performance may degrade. Our proposed Iterative Candidate Elimination (ICE) procedure makes the iterative ambiguity resolution possible by gradually eliminating a portion of least likely candidates in ambiguously labeled face. We further extend MCar to incorporate the labeling constraints between instances when such prior knowledge is available. Compared to existing methods, our approach demonstrates improvement on several ambiguously labeled datasets.
https://arxiv.org/abs/1702.04455
Do GANS (Generative Adversarial Nets) actually learn the target distribution? The foundational paper of (Goodfellow et al 2014) suggested they do, if they were given sufficiently large deep nets, sample size, and computation time. A recent theoretical analysis in Arora et al (to appear at ICML 2017) raised doubts whether the same holds when discriminator has finite size. It showed that the training objective can approach its optimum value even if the generated distribution has very low support —in other words, the training objective is unable to prevent mode collapse. The current note reports experiments suggesting that such problems are not merely theoretical. It presents empirical evidence that well-known GANs approaches do learn distributions of fairly low support, and thus presumably are not learning the target distribution. The main technical contribution is a new proposed test, based upon the famous birthday paradox, for estimating the support size of the generated distribution.
https://arxiv.org/abs/1706.08224
Using in situ grazing-incidence x-ray scattering, we have measured the diffuse scattering from islands that form during layer-by-layer growth of GaN by metal-organic vapor phase epitaxy on the (1010) m-plane surface. The diffuse scattering is extended in the (0001) in-plane direction in reciprocal space, indicating a strong anisotropy with islands elongated along [1 $\overline{2}$ 10] and closely spaced along [0001]. This is confirmed by atomic force microscopy of a quenched sample. Islands were characterized as a function of growth rate G and temperature. The island spacing along [0001] observed during the growth of the first monolayer obeys a power-law dependence on growth rate G$^{-n}$, with an exponent $n = 0.25 \pm 0.02$. Results are in agreement with recent kinetic Monte Carlo simulations, indicating that elongated islands result from the dominant anisotropy in step edge energy and not from surface diffusion anisotropy. The observed power-law exponent can be explained using a simple steady-state model, which gives n = 1/4.
https://arxiv.org/abs/1706.09955
Interest in neural machine translation has grown rapidly as its effectiveness has been demonstrated across language and data scenarios. New research regularly introduces architectural and algorithmic improvements that lead to significant gains over “vanilla” NMT implementations. However, these new techniques are rarely evaluated in the context of previously published techniques, specifically those that are widely used in state-of-theart production and shared-task systems. As a result, it is often difficult to determine whether improvements from research will carry over to systems deployed for real-world use. In this work, we recommend three specific methods that are relatively easy to implement and result in much stronger experimental systems. Beyond reporting significantly higher BLEU scores, we conduct an in-depth analysis of where improvements originate and what inherent weaknesses of basic NMT models are being addressed. We then compare the relative gains afforded by several other techniques proposed in the literature when starting with vanilla systems versus our stronger baselines, showing that experimental conclusions may change depending on the baseline chosen. This indicates that choosing a strong baseline is crucial for reporting reliable experimental results.
https://arxiv.org/abs/1706.09733
Optimizing the performance of GPU kernels is challenging for both human programmers and code generators. For example, CUDA programmers must set thread and block parameters for a kernel, but might not have the intuition to make a good choice. Similarly, compilers can generate working code, but may miss tuning opportunities by not targeting GPU models or performing code transformations. Although empirical autotuning addresses some of these challenges, it requires extensive experimentation and search for optimal code variants. This research presents an approach for tuning CUDA kernels based on static analysis that considers fine-grained code structure and the specific GPU architecture features. Notably, our approach does not require any program runs in order to discover near-optimal parameter settings. We demonstrate the applicability of our approach in enabling code autotuners such as Orio to produce competitive code variants comparable with empirical-based methods, without the high cost of experiments.
https://arxiv.org/abs/1701.08547
This paper presents a co-salient object detection method to find common salient regions in a set of images. We utilize deep saliency networks to transfer co-saliency prior knowledge and better capture high-level semantic information, and the resulting initial co-saliency maps are enhanced by seed propagation steps over an integrated graph. The deep saliency networks are trained in a supervised manner to avoid online weakly supervised learning and exploit them not only to extract high-level features but also to produce both intra- and inter-image saliency maps. Through a refinement step, the initial co-saliency maps can uniformly highlight co-salient regions and locate accurate object boundaries. To handle input image groups inconsistent in size, we propose to pool multi-regional descriptors including both within-segment and within-group information. In addition, the integrated multilayer graph is constructed to find the regions that the previous steps may not detect by seed propagation with low-level descriptors. In this work, we utilize the useful complementary components of high-, low-level information, and several learning-based steps. Our experiments have demonstrated that the proposed approach outperforms comparable co-saliency detection methods on widely used public databases and can also be directly applied to co-segmentation tasks.
https://arxiv.org/abs/1706.09650
A pencil-like morphology of homoepitaxially grown GaN nanowires is exploited for the fabrication of thin conformal intrawire InGaN nanoshells which host quantum dots in nonpolar, semipolar and polar crystal regions. All three quantum dot types exhibit single photon emission with narrow emission line widths and high degrees of linear optical polarization. The host crystal region strongly affects both single photon wavelength and emission lifetime, reaching subnanosecond time scales for the non- and semipolar quantum dots. Localization sites in the InGaN potential landscape, most likely induced by indium fluctuations across the InGaN nanoshell, are identified as the driving mechanism for the single photon emission. The hereby reported pencil-like InGaN nanoshell is the first single nanostructure able to host all three types of single photon sources and is, thus, a promising building block for tunable quantum light devices integrated into future photonic circuits.
https://arxiv.org/abs/1706.03603
Recommender systems aim to find an accurate and efficient mapping from historic data of user-preferred items to a new item that is to be liked by a user. Towards this goal, energy-based sequence generative adversarial nets (EB-SeqGANs) are adopted for recommendation by learning a generative model for the time series of user-preferred items. By recasting the energy function as the feature function, the proposed EB-SeqGANs is interpreted as an instance of maximum-entropy imitation learning.
https://arxiv.org/abs/1706.09200
Everything that exists in R is an object [Chambers2016]. This article examines what would be possible if we kept copies of all R objects that have ever been created. Not only objects but also their properties, meta-data, relations with other objects and information about context in which they were created. We introduce archivist, an R package designed to improve the management of results of data analysis. Key functionalities of this package include: (i) management of local and remote repositories which contain R objects and their meta-data (objects’ properties and relations between them); (ii) archiving R objects to repositories; (iii) sharing and retrieving objects (and it’s pedigree) by their unique hooks; (iv) searching for objects with specific properties or relations to other objects; (v) verification of object’s identity and context of it’s creation. The presented archivist package extends, in a combination with packages such as knitr and Sweave, the reproducible research paradigm by creating new ways to retrieve and validate previously calculated objects. These new features give a variety of opportunities such as: sharing R objects within reports or articles; adding hooks to R objects in table or figure captions; interactive exploration of object repositories; caching function calls with their results; retrieving object’s pedigree (information about how the object was created); automated tracking of the performance of considered models, restoring R libraries to the state in which object was archived.
https://arxiv.org/abs/1706.08822
In this paper, we investigate the Chinese calligraphy synthesis problem: synthesizing Chinese calligraphy images with specified style from standard font(eg. Hei font) images (Fig. 1(a)). Recent works mostly follow the stroke extraction and assemble pipeline which is complex in the process and limited by the effect of stroke extraction. We treat the calligraphy synthesis problem as an image-to-image translation problem and propose a deep neural network based model which can generate calligraphy images from standard font images directly. Besides, we also construct a large scale benchmark that contains various styles for Chinese calligraphy synthesis. We evaluate our method as well as some baseline methods on the proposed dataset, and the experimental results demonstrate the effectiveness of our proposed model.
https://arxiv.org/abs/1706.08789
Neural machine translation (NMT) has achieved notable performance recently. However, this approach has not been widely applied to the translation task between Chinese and Uyghur, partly due to the limited parallel data resource and the large proportion of rare words caused by the agglutinative nature of Uyghur. In this paper, we collect ~200,000 sentence pairs and show that with this middle-scale database, an attention-based NMT can perform very well on Chinese-Uyghur/Uyghur-Chinese translation. To tackle rare words, we propose a novel memory structure to assist the NMT inference. Our experiments demonstrated that the memory-augmented NMT (M-NMT) outperforms both the vanilla NMT and the phrase-based statistical machine translation (SMT). Interestingly, the memory structure provides an elegant way for dealing with words that are out of vocabulary.
https://arxiv.org/abs/1706.08683
We introduce a minimal model for the evolution of functional protein-interaction networks using a sequence-based mutational algorithm, and apply the model to study neutral drift in networks that yield oscillatory dynamics. Starting with a functional core module, random evolutionary drift increases network complexity even in the absence of specific selective pressures. Surprisingly, we uncover a hidden order in sequence space that gives rise to long-term evolutionary memory, implying strong constraints on network evolution due to the topology of accessible sequence space.
https://arxiv.org/abs/1706.08499
Generalized linear model with $L_1$ and $L_2$ regularization is a widely used technique for solving classification, class probability estimation and regression problems. With the numbers of both features and examples growing rapidly in the fields like text mining and clickstream data analysis parallelization and the use of cluster architectures becomes important. We present a novel algorithm for fitting regularized generalized linear models in the distributed environment. The algorithm splits data between nodes by features, uses coordinate descent on each node and line search to merge results globally. Convergence proof is provided. A modifications of the algorithm addresses slow node problem. For an important particular case of logistic regression we empirically compare our program with several state-of-the art approaches that rely on different algorithmic and data spitting methods. Experiments demonstrate that our approach is scalable and superior when training on large and sparse datasets.
https://arxiv.org/abs/1611.02101
The circumstellar environments of objects on the asymptotic giant branch and beyond are rich in molecular species. Nevertheless, methanol has never been detected in such an object, and is therefore often taken as a clear signpost for a young stellar object. However, we report the first detection of CH3OH in a post-AGB object, HD101584, using ALMA. Its emission, together with emissions from CO, SiO, SO, CS, and H2CO, comes from two extreme velocity spots on either side of the object where a high-velocity outflow appears to interact with the surrounding medium. We have derived molecular abundances, and propose that the detected molecular species are the effect of a post-shock chemistry where circumstellar grains play a role. We further provide evidence that HD101584 was a low-mass, M-type AGB star.
https://arxiv.org/abs/1706.08254
Neural machine translation (NMT) has recently become popular in the field of machine translation. However, NMT suffers from the problem of repeating or missing words in the translation. To address this problem, Tu et al. (2017) proposed an encoder-decoder-reconstructor framework for NMT using back-translation. In this method, they selected the best forward translation model in the same manner as Bahdanau et al. (2015), and then trained a bi-directional translation model as fine-tuning. Their experiments show that it offers significant improvement in BLEU scores in Chinese-English translation task. We confirm that our re-implementation also shows the same tendency and alleviates the problem of repeating and missing words in the translation on a English-Japanese task too. In addition, we evaluate the effectiveness of pre-training by comparing it with a jointly-trained model of forward translation and back-translation.
https://arxiv.org/abs/1706.08198
Semantic boundary and edge detection aims at simultaneously detecting object edge pixels in images and assigning class labels to them. Systematic training of predictors for this task requires the labeling of edges in images which is a particularly tedious task. We propose a novel strategy for solving this task, when pixel-level annotations are not available, performing it in an almost zero-shot manner by relying on conventional whole image neural net classifiers that were trained using large bounding boxes. Our method performs the following two steps at test time. Firstly it predicts the class labels by applying the trained whole image network to the test images. Secondly, it computes pixel-wise scores from the obtained predictions by applying backprop gradients as well as recent visualization algorithms such as deconvolution and layer-wise relevance propagation. We show that high pixel-wise scores are indicative for the location of semantic boundaries, which suggests that the semantic boundary problem can be approached without using edge labels during the training phase.
https://arxiv.org/abs/1606.09187
State-of-the-art approaches for image captioning require supervised training data consisting of captions with paired image data. These methods are typically unable to use unsupervised data such as textual data with no corresponding images, which is a much more abundant commodity. We here propose a novel way of using such textual data by artificially generating missing visual information. We evaluate this learning approach on a newly designed model that detects visual concepts present in an image and feed them to a reviewer-decoder architecture with an attention mechanism. Unlike previous approaches that encode visual concepts using word embeddings, we instead suggest using regional image features which capture more intrinsic information. The main benefit of this architecture is that it synthesizes meaningful thought vectors that capture salient image properties and then applies a soft attentive decoder to decode the thought vectors and generate image captions. We evaluate our model on both Microsoft COCO and Flickr30K datasets and demonstrate that this model combined with our semi-supervised learning method can largely improve performance and help the model to generate more accurate and diverse captions.
https://arxiv.org/abs/1611.05321