Publications

Publications
Publications
We strongly believe in open source and giving to our community. We work directly with researchers in academia and seek out new perspectives with our intern and fellowship programs. We generalize our solutions and release them to the world as open source projects. We host discussions and publish our results.

Publications

arXiv, June, 2014

When relevance is not Enough: Promoting Visual Attractiveness for Fashion E-commerce

Wei Di, Anurag Bhardwaj, Vignesh Jagadeesh, Robinson Piramuthu, Elizabeth Churchill

Fashion, and especially apparel, is the fastest-growing category in online shopping. As consumers requires sensory experience especially for apparel goods for which their appearance matters most, images play a key role not only in conveying crucial information that is hard to express in text, but also in affecting consumer's attitude and emotion towards the product.

However, research related to e-commerce product image has mostly focused on quality at perceptual level, but not the quality of content, and the way of presenting.This study aims to address the effectiveness of types of image in showcasing fashion apparel in terms of its attractiveness, i.e. the ability to draw consumer's attention, interest, and in return their engagement.

We apply advanced vision technique to quantize attractiveness using three common display types in fashion filed, i.e. human model, mannequin, and flat. We perform two-stage study by starting with large scale behavior data from real online market, then moving to well designed user experiment to further deepen our understandings on consumer's reasoning logic behind the action.

We propose a Fisher noncentral hypergeometric distribution based user choice model to quantitatively evaluate user's preference. Further, we investigate the potentials to leverage visual impact for a better search that caters to user's preference. A visual attractiveness based re-ranking model that incorporates both presentation efficacy and user preference is proposed. We show quantitative improvement by promoting visual attractiveness into search on top of relevance.

Keywords
Categories
arXiv, May, 2014

Enhancing Visual Fashion Recommendations with Users in the Loop

Anurag Bhardwaj, Vignesh Jagadeesh, Wei Di, Robinson Piramuthu, Elizabeth Churchill

We describe a completely automated large scale visual recommendation system for fashion. Existing approaches have primarily relied on purely computational models to solving this problem that ignore the role of users in the system.

In this paper, we propose to overcome this limitation by incorporating a user-centric design of visual fashion recommendations. Specifically, we propose a technique that augments 'user preferences' in models by exploiting elasticity in fashion choices. We further design a user study on these choices and gather results from the 'wisdom of crowd' for deeper analysis.

Our key insights learnt through these results suggest that fashion preferences when constrained to a particular class, contain important behavioral signals that are often ignored in recommendation design.

Further, presence of such classes also reflect strong correlations to visual perception which can be utilized to provide aesthetically pleasing user experiences. Finally, we illustrate that user approval of visual fashion recommendations can be substantially improved by carefully incorporating these user-centric feedback into the system framework.

CVPR 2014

Region-based Discriminative Feature Pooling for Scene Text Recognition

Chen Yu Lee, Anurag Bhardwaj, Wei Di, Vignesh Jagadeesh, Robinson Piramuthu

We present a new feature representation method for scene text recognition problem, particularly focusing on improving scene character recognition. Many existing methods rely on Histogram of Oriented Gradient (HOG) or part based models, which do not span the feature space well for characters in natural scene images, especially given large variation in fonts with cluttered backgrounds.

In this work, we propose a discriminative feature pooling method that automatically learns the most informative sub-regions of each scene character within a multi-class classification framework, whereas each sub-region seamlessly integrates a set of low-level image features through integral images.

The proposed feature representation is compact, computationally efficient, and able to effectively model distinctive spatial structures of each individual character class. Extensive experiments conducted on challenging datasets (Chars74K, ICDAR’03, ICDAR’11, SVT) show that our method significantly outperforms existing methods on scene character classification and scene text recognition tasks.

Keywords
Categories
Tutorial at WWW-2014

E-commerce Product Search: Personalization, Diversification, and beyond

Atish Das Sarma, Nish Parikh, Neel Sundaresan

The focus of this tutorial will be e-commerce product search. Several research challenges appear in this context, both from a research standpoint as well as an application standpoint. We will present various approaches adopted in the industry,

review well-known research techniques developed over the last decade, draw parallels to traditional web search highlighting the new challenges in this setting, and dig deep into some of the algorithmic and technical approaches developed for this context.

A specific approach that will involve a deep dive into literature, theoretical techniques, and practical impact is that of identifying most suited results quickly from a large database, with settings various across cold start users, and those for whom personalization is possible.

In this context, top-k and skylines will be discussed specifically as they form a key approach that spans the web, data mining, and database communities and presents a powerful tool for search across multi-dimensional items with clear preferences within each attribute, like product search as opposed to regular web search.

Keywords
ICML 2014 workshop on New Learning Models and Frameworks for BigData

Geometric VLAD for Large Scale Image Search

Zixuan Wang, Wei Di, Anurag Bhardwaj, Vignesh Jagadeesh, Robinson Piramuthu

We present a novel compact image descriptor for large scale image search. Our proposed descriptor - Geometric VLAD (gVLAD) is an extension of VLAD (Vector of locally Aggregated Descriptors) that incorporates weak geometry information into the VLAD framework.

The proposed geometry cues are derived as a membership function over keypoint angles which contain evident and informative information but yet often discarded. A principled technique for learning the membership function by clustering angles is also presented.

Further, to address the overhead of iterative codebook training over real-time datasets, a novel codebook adaptation strategy is outlined. Finally, we demonstrate the efficacy of proposed gVLAD based retrieval framework where we achieve more than 15% improvement in mAP over existing benchmarks.

Keywords
Categories
IEEE International Conference on Image Processing (ICIP), 2014

Cascaded Sparse Color-Localized Matching for Logo Retrieval

Rohit Pandey, Wei Di, Vignesh Jagadeesh, Robinson Piramuthu, Anurag Bhardwaj

We present a new feature representation method for scene text recognition problem, particularly focusing on improving scene character recognition. Many existing methods rely on Histogram of Oriented Gradient (HOG) or part-based models, which do not span the feature space well for characters in natural scene images, especially given large variation in fonts with cluttered backgrounds.

In this work, we propose a discriminative feature pooling method that automatically learns the most informative sub-regions of each scene character within a multi-class classification framework, whereas each sub-region seamlessly integrates a set of low-level image features through integral images.

The proposed feature representation is compact, computationally efficient, and able to effectively model distinctive spatial structures of each individual character class. Extensive experiments conducted on challenging datasets (Chars74K, ICDAR’03, ICDAR’11, SVT) show that our method significantly outperforms existing methods on scene character classification and scene text recognition tasks.

Keywords
Categories
Journal of the Association for Computing Machinery (JACM) – 2013

Distributed Random Walks

Atish Das Sarma, Danupon Nanongkai, Gopal Pandurangan, Prasad Tetali

Performing random walks in networks is a fundamental primitive that has found applications in many areas of computer science, including distributed computing. In this article, we focus on the problem of sampling random walks efficiently in a distributed network and its applications. Given bandwidth constraints, the goal is to minimize the number of rounds required to obtain random walk samples. All previous algorithms that compute a random walk sample of length ℓ as a subroutine always do so naively, that is, in O(ℓ) rounds.

The main contribution of this article is a fast distributed algorithm for performing random walks. We present a sublinear time distributed algorithm for performing random walks whose time complexity is sublinear in the length of the walk. Our algorithm performs a random walk of length ℓ in Õ(√ℓD) rounds (Õ hides polylog n factors where n is the number of nodes in the network) with high probability on an undirected network, where D is the diameter of the network. For small diameter graphs, this is a significant improvement over the naive O(ℓ) bound.

Furthermore, our algorithm is optimal within a poly-logarithmic factor as there exists a matching lower bound [Nanongkai et al. 2011]. We further extend our algorithms to efficiently perform k independent random walks in Õ(√kℓD + k) rounds. We also show that our algorithm can be applied to speedup the more general Metropolis-Hastings sampling. Our random-walk algorithms can be used to speed up distributed algorithms in applications that use random walks as a subroutine. We present two main applications.

First, we give a fast distributed algorithm for computing a random spanning tree (RST) in an arbitrary (undirected unweighted) network which runs in Õ(√mD) rounds with high probability (m is the number of edges). Our second application is a fast decentralized algorithm for estimating mixing time and related parameters of the underlying network. Our algorithm is fully decentralized and can serve as a building block in the design of topologically-aware networks.

Keywords

Pages