Pinterest Image Search Engine helps hundreds of millions of users discover interesting content everyday. This motivates us to improve the image search quality by evolving our ranking techniques. In this work, we share how we practically design and deploy various ranking pipelines into Pinterest image search ecosystem. Specifically, we focus on introducing our novel research and study on three aspects: training data, user/image featurization and ranking models. Extensive offline and online studies compared the performance of different models and demonstrated the efficiency and effectiveness of our final launched ranking models.

Linhong Zhu

We present a neural network framework for learning a survival model to predict a time-to-event outcome while simultaneously learning a topic model that reveals feature relationships. In particular, we model each subject as a distribution over "topics", where a topic could, for instance, correspond to an age group, a disorder, or a disease. The presence of a topic in a subject means that specific clinical features are more likely to appear for the subject. Topics encode information about related features and are learned in a supervised manner to predict a time-to-event outcome. Our framework supports combining many different topic and survival models; training the resulting joint survival-topic model readily scales to large datasets using standard neural net optimizers with minibatch gradient descent. For example, a special case is to combine LDA with a Cox model, in which case a subject's distribution over topics serves as the input feature vector to the Cox model. We explain how to address practical implementation issues that arise when applying these neural survival-supervised topic models to clinical data, including how to visualize results to assist clinical interpretation. We study the effectiveness of our proposed framework on seven clinical datasets on predicting time until death as well as hospital ICU length of stay, where we find that neural survival-supervised topic models achieve competitive accuracy with existing approaches while yielding interpretable clinical topics that explain feature relationships. Our code is available at: https://github.com/georgehc/survival-topics

George H. Chen Linhong Li Ren Zuo Amanda Coston Jeremy C. Weiss

We compute higher Frobenius-Schur indicators of pq-dimensional pointed Hopf algebras in characteristic p through their associated graded Hopf algebras. These indicators are gauge invariants for the monoidal categories of representations of these algebras.

Si Chen Tiantian Liu Linhong Wang Xingting Wang

We present a silicon-photonic tensor core using 2D ferroelectric materials to enable wavelength- and polarization-domain computing. Results, based on experimentally characterized material properties, show up to 83% improvement in computation accuracy compared to coherent networks.

Amin Shafiee Linhong Chen Sudeep Pasricha Jie Yao Mahdi Nikdast

In general the endomorphisms of a non-abelian group do not form a ring under the operations of addition and composition of functions. Several papers have dealt with the ring of functions defined on a group which are endomorphisms when restricted to the elements of a cover of the group by abelian subgroups. We give an algorithm which allows us to determine the elements of the ring of functions of a finite $p$-group which arises in this manner when the elements of the cover are required to be either cyclic or elementary abelian of rank $2$. This enables us to determine the actual structure of such a ring as a subdirect product. A key part of the argument is the construction of a graph whose vertices are the subgroups of order $p$ and whose edges are determined by the covering.

Gary Walls Linhong Wang

The growing popularity of social media (e.g, Twitter) allows users to easily share information with each other and influence others by expressing their own sentiments on various subjects. In this work, we propose an unsupervised \emph{tri-clustering} framework, which analyzes both user-level and tweet-level sentiments through co-clustering of a tripartite graph. A compelling feature of the proposed framework is that the quality of sentiment clustering of tweets, users, and features can be mutually improved by joint clustering. We further investigate the evolution of user-level sentiments and latent feature vectors in an online framework and devise an efficient online algorithm to sequentially update the clustering of tweets, users and features with newly arrived data. The online framework not only provides better quality of both dynamic user-level and tweet-level sentiment analysis, but also improves the computational and storage efficiency. We verified the effectiveness and efficiency of the proposed approaches on the November 2012 California ballot Twitter data.

Linhong Zhu Aram Galstyan James Cheng Kristina Lerman

Social media can be viewed as a social system where the currency is attention. People post content and interact with others to attract attention and gain new followers. In this paper, we examine the distribution of attention across a large sample of users of a popular social media site Twitter. Through empirical analysis of these data we conclude that attention is very unequally distributed: the top 20\% of Twitter users own more than 96\% of all followers, 93\% of the retweets, and 93\% of the mentions. We investigate the mechanisms that lead to attention inequality and find that it results from the "rich-get-richer" and "poor-get-poorer" dynamics of attention diffusion. Namely, users who are "rich" in attention, because they are often mentioned and retweeted, are more likely to gain new followers, while those who are "poor" in attention are likely to lose followers. We develop a phenomenological model that quantifies attention diffusion and network dynamics, and solve it to study how attention inequality grows over time in a dynamic environment of social media.

Linhong Zhu Kristina Lerman

In this paper, for the first time, we study label propagation in heterogeneous graphs under heterophily assumption. Homophily label propagation (i.e., two connected nodes share similar labels) in homogeneous graph (with same types of vertices and relations) has been extensively studied before. Unfortunately, real-life networks are heterogeneous, they contain different types of vertices (e.g., users, images, texts) and relations (e.g., friendships, co-tagging) and allow for each node to propagate both the same and opposite copy of labels to its neighbors. We propose a $\mathcal{K}$-partite label propagation model to handle the mystifying combination of heterogeneous nodes/relations and heterophily propagation. With this model, we develop a novel label inference algorithm framework with update rules in near-linear time complexity. Since real networks change over time, we devise an incremental approach, which supports fast updates for both new data and evidence (e.g., ground truth labels) with guaranteed efficiency. We further provide a utility function to automatically determine whether an incremental or a re-modeling approach is favored. Extensive experiments on real datasets have verified the effectiveness and efficiency of our approach, and its superiority over the state-of-the-art label propagation methods.

Dingxiong Deng Fan Bai Yiqi Tang Shuigeng Zhou Cyrus Shahabi Linhong Zhu

We study skew inverse power series extensions R[[y^{-1};tau,delta]], where R is a noetherian ring equipped with an automorphism tau and a tau-derivation delta. We find that these extensions share many of the well known features of commutative power series rings. As an application of our analysis, we see that the iterated skew inverse power series rings corresponding to nth Weyl algebras are complete local, noetherian, Auslander regular domains whose right Krull dimension, global dimension, and classical Krull dimension, are all equal to 2n.

Edward S. Letzter Linhong Wang

Let A be a semprime, right noetherian ring equipped with an automorphism alpha, and let B := A[[y; alpha]] denote the corresponding skew power series ring (which is also semiprime and right noetherian). We prove that the Goldie ranks of A and B are equal. We also record applications to induced ideals.

Edward S. Letzter Linhong Wang