Abstract Traditional query auto-completion (QAC) relies heavily on search logs collected over many users. However, in on-device email search, the scarcity of logs and the governing privacy constraints make QAC a challenging task. In this work, we propose an on-device QAC method that runs directly on users’ devices, where users’ sensitive data and interaction logs are not collected, shared, or aggregated through web services. This method retrieves candidates using pseudo relevance feedback, and ranks them based on relevance signals that explore the textual and structural information from users’ emails. We also propose a private corpora based evaluation method, and empirically demonstrate the effectiveness of our proposed method.

GNSS coordinate time series reflects the combined influence of geophysical factors on stations around the land surface. Although some traditional parameterized methods are helpful to determine the magnitude of the seasonal signal at GNSS stations, the annual variation characteristics of the stations are not static, thus it is quite necessary to extract finer periodic signals with time-varying coefficients (PSTC) from stations’ position time series. This paper focuses on the height time series of 243 stations from the Crustal Movement Observation Network of China (CMONOC) and employs singular spectrum analysis (SSA) to extract PSTC. The results show that SSA method can effectively extract the time-varying trend and periodic terms from the original time series, which cannot be perfectly achieved by parameterized methods. SSA method reduces the RMSE value of the residual time series at 90.5% CMONOC stations, compared with the results of maximum likelihood estimation (MLE). Its function in extracting the PSTC from CMONOC stations is significant for further explaining the generation mechanism of the land surface nonlinear deformation in China. Different from MLE method which only considers the given epochs of offsets, SSA method can effectively fit the original time series through singular value decomposition (SVD) and signal reconstruction, despite there are unrecognized offsets contained in GNSS time series. It still works well when there is an offset up to 20 mm, which would reduce the traditional workload of offset detection by sight. SSA method manages to distinguish large unknown offsets, showing as negative improvement rates.

Machine learning methods are used to discover complex nonlinear relationships in biological and medical data. However, sophisticated learning models are computationally unfeasible for data with millions of features. Here, we introduce the first feature selection method for nonlinear learning problems that can scale up to large, ultra-high dimensional biological data. More specifically, we scale up the novel Hilbert-Schmidt Independence Criterion Lasso (HSIC Lasso) to handle millions of features with tens of thousand samples. The proposed method is guaranteed to find an optimal subset of maximally predictive features with minimal redundancy, yielding higher predictive power and improved interpretability. Its effectiveness is demonstrated through applications to classify phenotypes based on module expression in human prostate cancer patients and to detect enzymes among protein structures. We achieve high accuracy with as few as 20 out of one million features-a dimensionality reduction of 99.998 percent. Our algorithm can be implemented on commodity cloud computing platforms. The dramatic reduction of features may lead to the ubiquitous deployment of sophisticated prediction models in mobile health care applications.

Multi-Armed Bandit (MAB) framework has been successfully applied in many web applications. However, many complex real-world applications that involve multiple content recommendations cannot fit into the traditional MAB setting. To address this issue, we consider an ordered combinatorial semi-bandit problem where the learner recommends S actions from a base set of K actions, and displays the results in S (out of M) different positions. The aim is to maximize the cumulative reward with respect to the best possible subset and positions in hindsight. By the adaptation of a minimum-cost maximum-flow network, a practical algorithm based on Thompson sampling is derived for the (contextual) combinatorial problem, thus resolving the problem of computational intractability.With its potential to work with whole-page recommendation and any probabilistic models, to illustrate the effectiveness of our method, we focus on Gaussian process optimization and a contextual setting where click-through rate is predicted using logistic regression. We demonstrate the algorithms’ performance on synthetic Gaussian process problems and on large-scale news article recommendation datasets from Yahoo! Front Page Today Module.

We propose a new random pruning method (called “submodular sparsification (SS)”) to reduce the cost of submodular maximization. The pruning is applied via a “submodularity graph” over the $n$ ground elements, where each directed edge is associated with a pairwise dependency defined by the submodular function. In each step, SS prunes a $1-1/\sqrt{c}$ (for $c>1$) fraction of the nodes using weights on edges computed based on only a small number ($O(\log n)$) of randomly sampled nodes. The algorithm requires $\log_\sqrt{c}n$ steps with a small and highly parallelizable per-step computation. An accuracy-speed tradeoff parameter $c$, set as $c = 8$, leads to a fast shrink rate $\sqrt{2}/4$ and small iteration complexity $\log_{2\sqrt{2}}n$. Analysis shows that w.h.p., the greedy algorithm on the pruned set of size $O(\log^2 n)$ can achieve a guarantee similar to that of processing the original dataset. In news and video summarization tasks, SS is able to substantially reduce both computational costs and memory usage, while maintaining (or even slightly exceeding) the quality of the original (and much more costly) greedy algorithm.

Query auto-completion (QAC) is widely used by modern search engines to assist users by predicting their intended queries. Most QAC approaches rely on deterministic batch learning algorithms trained from past query log data. However, query popularities keep changing all the time and QAC operates in a real-time scenario where users interact with the search engine continually. So, ideally, QAC must be timely and adaptive enough to reflect time-sensitive changes in an online fashion. Second, due to the vertical position bias, a query suggestion with a higher rank tends to attract more clicks regardless of user's original intention. Hence, in the long run, it is important to place some lower ranked yet potentially more relevant queries to higher positions to collect more valuable user feedbacks. In order to tackle these issues, we propose to formulate QAC as a ranked Multi-Armed Bandits (MAB) problem which enjoys theoretical soundness. To utilize prior knowledge from query logs, we propose to use Bayesian inference and Thompson Sampling to solve this MAB problem. Extensive experiments on large scale datasets show that our QAC algorithm has the capacity to adaptively learn temporal trends, and outperforms existing QAC algorithms in ranking qualities.

It is widely known that there exists a semantic gap between web documents and user queries and bridging this gap is crucial to advance information retrieval systems. The task of query rewriting, aiming to alter a given query to a rewrite query that can close the gap and improve information retrieval performance, has attracted increasing attention in recent years. However, the majority of existing query rewriters are not designed to boost search performance and consequently their rewrite queries could be sub-optimal. In this paper, we propose a learning to rewrite framework that consists of a candidate generating phase and a candidate ranking phase. The candidate generating phase provides us the flexibility to reuse most of existing query rewriters; while the candidate ranking phase allows us to explicitly optimize search relevance. Experimental results on a commercial search engine demonstrate the effectiveness of the proposed framework. Further experiments are conducted to understand the important components of the proposed framework.

Search engines play a crucial role in our daily lives. Relevance is the core problem of a commercial search engine. It has attracted thousands of researchers from both academia and industry and has been studied for decades. Relevance in a modern search engine has gone far beyond text matching, and now involves tremendous challenges. The semantic gap between queries and URLs is the main barrier for improving base relevance. Clicks help provide hints to improve relevance, but unfortunately for most tail queries, the click information is too sparse, noisy, or missing entirely. For comprehensive relevance, the recency and location sensitivity of results is also critical. In this paper, we give an overview of the solutions for relevance in the Yahoo search engine. We introduce three key techniques for base relevance -- ranking functions, semantic matching features and query rewriting. We also describe solutions for recency sensitive relevance and location sensitive relevance. This work builds upon 20 years of existing efforts on Yahoo search, summarizes the most recent advances and provides a series of practical relevance solutions. The performance reported is based on Yahoo's commercial search engine, where tens of billions of urls are indexed and served by the ranking system.

The Alternating Direction Method of Multipliers (ADMM) has received lots of attention recently due to the tremendous demand from large-scale and data-distributed machine learning applications. In this paper, we present a stochastic setting for optimization problems with non-smooth composite objective functions. To solve this problem, we propose a stochastic ADMM algorithm. Our algorithm applies to a more general class of convex and nonsmooth objective functions, beyond the smooth and separable least squares loss used in lasso. We also demonstrate the rates of convergence for our algorithm under various structural assumptions of the stochastic function: O(1/\sqrtt) for convex functions and O(\log t/t) for strongly convex functions. Compared to previous literature, we establish the convergence rate of ADMM for convex problems in terms of both the objective value and the feasibility violation. A novel application named Graph-Guided SVM is proposed to demonstrate the usefulness of our algorithm.

The classic Stochastic Approximation (SA) method achieves optimal rates under the black-box model. This optimality does not rule out better algorithms when more information about functions and data is available.

This work aimed to refine the Moderate Resolution Imaging Spectroradiometer (MODIS) based snow cover algorithm for the Hindu Kush-Himalayan (HKH) region. Taking into account the effect of the atmosphere and terrain on the satellite observations at the top of the atmosphere (TOA), particularly in heavily rugged Tibet plateau region, the surface reflectances were retrieved from the TOA reflectances after atmospheric and topographic corrections. To reduce the effects of the snow/cloud confusion, a normalized difference cloud index (NDCI) model was proposed to discriminate snow/cloud pixels, apart from use of the MODIS cloud mask product MOD35. Furthermore, MODIS land surface temperature (LST) product MOD11_L2 have been used to ensure better accuracy of the snow cover pixels. Comparisons of the resultant MODIS snow cover with those obtained respectively from high resolution Landsat ETM+ data and the MODIS snow cover product MOD10_L2 for the Mount Everest region at different seasons, showed overestimation of the MOD10_L2 snow cover with the differences of 50%, whereas the improved algorithm can estimate the snow cover for HKH region more precisely with absolute accuracy of 90%.

The high computational cost of nonlinear support vector machines has limited their usability for large-scale problems. We propose two novel stochastic algorithms to tackle this problem. These algorithms are based on a simple and classic optimization method: the Frank-Wolfe method, which is known to be fast for problems with a large number of linear constraints. Formulating the nonlinear SVM problem to take advantage of this method, we achieve a provable time complexity of O(dQ2/∈2). The proposed algorithms achieve comparable or even better accuracies than the state-of-the-art methods, and are significantly faster.

The long-standing problem of efficient nearest-neighbor (NN) search has ubiquitous applications ranging from astrophysics to MP3 fingerprinting to bioinformatics to movie recommendations. As the dimensionality of the dataset increases, exact NN search becomes computationally prohibitive; (1+eps)-distance-approximate NN search can provide large speedups but risks losing the meaning of NN search present in the ranks (ordering) of the distances. This paper presents a simple, practical algorithm allowing the user to, for the first time, directly control the true accuracy of NN search (in terms of ranks) while still achieving the large speedups over exact NN. Experiments with high-dimensional datasets show that it often achieves faster and more accurate results than the best-known distance-approximate method, with much more stable behavior.

We consider the problem of learning dissimilarities between points via formulations which preserve a specified ordering between points rather than the numerical values of the dissimilarities. Dissimilarity ranking (d-ranking) learns from instances like "A is more similar to B than C is to D" or "The distance between E and F is larger than that between G and H". Three formulations of d-ranking problems are presented and new algorithms are presented for two of them, one by semidefinite programming (SDP) and one by quadratic programming (QP). Among the novel capabilities of these approaches are out-of-sample prediction and scalability to large problems.

This paper addresses the problem of speaker segmentation in telephone conversation. The segmentation is done in three steps: 1) preliminary segmentation to hypothesize speaker turning points; 2) clustering of segments; and 3) re-segmentation to determine speaker identity of each segment. It is found that vocal source related features are more speaker-discriminative than the conventional vocal tract related features for small amount of data. This motivates us to thoughtfully incorporate vocal source features into early stages of the speaker segmentation process, where decisions have to be made with limited data. Speaker segmentation experiments are carried out on 36 summed channel conversations in the NIST 2004 Speaker Recognition Evaluation. The proposed use of vocal source features leads to noticeable performance improvement

As the low-cost video transmission becomes popular, video-based bi-modal (audio and visual) authentication has great potential in various applications that require access control over handheld terminals. In this paper, we propose to use the averaged mouth image (AMI) for speaker verification. The AMI is computed by averaging properly aligned mouth images over the whole video sequence. Despite its simplicity, the AMI not only contains appearance information but also describes stylistic articulation gestures of individual speakers. The AMI is found to be fairly invariant against the spoken content. The experimental results show that the AMI based features are very effective in discriminating speaking persons. Explicit and precise extraction of lip contours or other feature points are not required. For bi-modal verification, the proposed video features are found to be highly complementary to the audio features

Afforestation in China at the nation-scale has been promoted by the government since 1999 with the provision of subsidies to buy environmental amenities. The article concerned with the impact on local communities according to household survey data in Dunhua County, Northeast China. Forty-two of the families involved in reforestation led a worse life after the program. In particular, it was important to understand what types of families were adversely affected and what kind of factors contributed to changes in the farmers' quality of life. It was found that 58.3% of the farmers perceived that converting cropland to forest was a volunteer action, and 39% of the farmers were worried about their future after 8 years of government subsidies. The survey showed that there was an increase of non-farm income and the economic crop income in Dunhua County. This implied that some of the farmers could find alternative income resources so that the household could remain high income even when they no longer receive program subsidies. However, it is important to note that some farmers led a worse life after reforestation. When the subsidy ended, 16% of the farmers plan to re-cultivate the land. Their intention clearly has implications for the sustainability of the program's environmental benefits. It was suggested that the government should pay more attention on the quality of reforestation instead of the ambitious scale. The officials should enhance public education programs and increase public participations in the decision-making process, which could improve the acceptance of land use change. In addition, the government should increase alternative economic crop or non-agriculture opportunities to assure the program's sustainability.