Abstract

Ideally a metric evaluating an abstract system summary should represent the extent to which the system-generated summary approximates the semantic inference conceived by the reader using a human-written reference summary. Most of the previous approaches relied upon word or syntactic sub-sequence overlap to evaluate system-generated summaries. Such metrics cannot evaluate the summary at semantic inference level. Through this work we introduce the metric of Semantic Similarity for Abstractive Summarization (SSAS), which leverages natural language inference and paraphrasing techniques to frame a novel approach to evaluate system summaries at semantic inference level. SSAS is based upon a weighted composition of quantities representing the level of agreement, contradiction, independence, paraphrasing, and optionally ROUGE score between a system-generated and a human-written summary.

Abstract

Extractive text summarization has been an extensive research problem in the field of natural language understanding. While the conventional approaches rely mostly on manually compiled features to generate the summary, few attempts have been made in developing data-driven systems for extractive summarization. To this end, we present a fully data-driven end-to-end deep network which we call as Hybrid MemNet for single document summarization task. The network learns the continuous unified representation of a document before generating its summary. It jointly captures local and global sentential information along with the notion of summary worthy sentences. Experimental results on two different corpora confirm that our model shows significant performance gains compared with the state-of-the-art baselines

Abstract

One of the most important and challenging problems in recommendation systems is that of modeling temporal behavior. Typically, modeling temporal behavior increases the cost of parameter inference and estimation. Along with it, it also poses the constraint of requiring a large amount of data for reliably learning the parameters of the model. Therefore, it is often difficult to model temporal behavior in large-scale real-world recommendation systems. In this work, we propose a deep neural network architecture which is based on a two level approach. We first generate document embeddings for every news article. We then use these embeddings and the previously read articles by a user to come up with her user profile. We then use this profile along with adequate positive and negative samples in order to train our model. The resulting model is then applied to a real-world data set. We compare it with a set of established baselines and the experimental results show that our model outperforms the state-of-the-art. We also use the learned model to recommend articles to users who have had very little interaction with items, i.e., have read a very less amount of news articles. We then demonstrate the effectiveness of our model to solve the problem of item cold-start.

Abstract

Deep neural networks have yielded immense success in speech recognition, computer vision and natural language processing. However, the exploration of deep neural networks for content based recommendation has received a relatively less amount of inspection. Also, different recommendation scenarios have their own issues which creates the need for different approaches for recommendation. One of the problems with news recommendation is that of handling temporal changes in user interests. Hence, modelling temporal behaviour in the domain of news recommendation becomes very important. In this work, we propose a recommendation model which uses semantic similarity between words as input to a 3-D Convolutional Neural Network in order to extract the temporal news reading pattern of the users. This in turn improves the quality of recommendations. We compare our model to a set of established baselines and the experimental results show that our model performs better than the state-of-the-art by 5.8% (Hit Ratio@10).

Abstract

It is very crucial for news aggregator websites which are recent in the market to actively engage its existing users. A recommendation system would help to tackle such a problem. However, due to the lack of sufficient amount of data, most of the state-of-the-art methods perform poorly in terms of recommending relevant news items to the users. In this paper, we propose a novel approach for Item-based Collaborative filtering for recommending news items using Markov Decision Process (MDP). Due to the sequential nature of news reading, we choose MDP to model our recommendation system as it is based on a sequence optimization paradigm. Further, we also incorporate factors like article freshness and similarity into our system by extrinsically modelling it in terms of reward for the MDP. We compare it with various other state-of-the-art methods. On a moderately low amount of data we see that our MDP-based approach outperforms the other approaches. One of the reasons for this is that the baselines fail to identify the underlying patterns within the sequence in which the articles are read by the users. Hence, the baselines are not able to generalize well.

Abstract

Social media is a rich source of information and opinion, with exponential data growth rate. However social media posts are difficult to analyze since they are brief, unstructured and noisy. Interestingly, many social media posts are about an entity or entities. Understanding which entity is central (Salient Entity) to a post, helps better analyze the post. In this paper we propose a model that aids in such analysis by identifying the Salient Entity in a social media post, tweets in particular. We present a supervised machine-learning model, to identify Salient Entity in a tweet and propose that the tweet is most likely about that particular entity. We have used the premise that, when an image accompanies a text, the text most likely is about the entity in that image, to build a dataset of tweets and salient entities. We trained our model using this dataset. Note that this does not restrict the applicability of our model in any way. We use tweets with images only to obtain objective ground truth data, while features for the model are derived from tweet text. Our experiments show that the model identifies Salient Named Entity with an F-measure of 0.63. We show the effectiveness of the proposed model for tweet-filtering and salience identification tasks. We have made the human annotated dataset and the source code of this model publicly available.

Abstract

Depth from focus is a highly accessible method to estimate the 3D structure of everyday scenes. Today’s DSLR and mobile cameras facilitate the easy capture of multiple focused images of a scene. Focus measures (FMs) that estimate the amount of focus at each pixel form the basis of depth-from-focus methods. Several FMs have been proposed in the past and new ones will emerge in the future, each with their own strengths. We estimate a weighted combination of standard FMs that outperforms others on a wide range of scene types. The resulting composite focus measure consists of FMs that are in consensus with one another but not in chorus. Our two-stage pipeline first estimates fine depth at each pixel using the composite focus measure. A cost-volume propagation step then assigns depths from confident pixels to others. We can generate high quality depth maps using just the top five FMs from our composite focus measure. This is a positive step towards depth estimation of everyday scenes with no special equipment.

Abstract

Smartphones have become the de-facto capture devices for everyday photography. Unlike traditional digital cameras, smartphones are versatile devices with auxiliary sensors, processing power, and networking capabilities. In this work, we harness the communication capabilities of smartphones and present a synchronous/co-ordinated multi-camera capture system. Synchronous capture is important for many image/video fusion and 3D reconstruction applications. The proposed system provides an inexpensive and effective means to capture multi-camera media for such applications. Our coordinated capture system is based on a wireless protocol that uses NTP based synchronization and device specific lag compensation. It achieves sub-frame synchronization across all participating smartphones of even heterogeneous make and model. We propose a new method based on fiducial markers displayed on an LCD screen to temporally calibrate smartphone cameras. We demonstrate the utility and versatility of this system to enhance traditional videography and to create novel visual representations such as panoramic videos, HDR videos, multi-view 3D reconstruction, multi-flash imaging, and multi-camera social media.

Abstract

View-graph is an essential input to large-scale structure from motion (SfM) pipelines. Accuracy and efficiency of large-scale SfM is crucially dependent on the input viewgraph. Inconsistent or inaccurate edges can lead to inferior or wrong reconstruction. Most SfM methods remove ‘undesirable’ images and pairs using several, fixed heuristic criteria, and propose tailor-made solutions to achieve specific reconstruction objectives such as efficiency, accuracy, or disambiguation. In contrast to these disparate solutions, we propose a single optimization framework that can be used to achieve these different reconstruction objectives with task-specific cost modeling. We also construct a very efficient network-flow based formulation for its approximate solution. The abstraction brought on by this selection mechanism separates the challenges specific to datasets and reconstruction objectives from the standard SfM pipeline and improves its generalization. This paper demonstrates the application of the proposed view-graph framework with standard SfM pipeline for two particular use-cases, (i) accurate and ghost-free reconstructions of highly ambiguous datasets using costs based on disambiguation priors, and (ii) accurate and efficient reconstruction of large-scale Internet datasets using costs based on commonly used priors.

Abstract

A class of recent approaches for generating images, called Generative Adversarial Networks (GAN), have been used to generate impressively realistic images of objects, bedrooms, handwritten digits and a variety of other image modalities. However, typical GAN-based approaches require large amounts of training data to capture the diversity across the image modality. In this paper, we propose DeLiGAN--a novel GAN-based architecture for diverse and limited training data scenarios. In our approach, we reparameterize the latent generative space as a mixture model and learn the mixture model's parameters along with those of GAN. This seemingly simple modification to the GAN framework is surprisingly effective and results in models which enable diversity in generated samples although trained with limited data. In our work, we show that DeLiGAN can generate images of handwritten digits, objects and hand-drawn sketches, all using limited amounts of data. To quantitatively characterize intra-class diversity of generated samples, we also introduce a modified version of" inception-score", a measure which has been found to correlate well with human assessment of generated samples