Personalized retrieval seeks to retrieve items relevant to a user event (e.g. a page visit or a query) that are adapted to the user's personal preferences. For example, two users who happen to perform the same event such as visiting the same product page or asking the same query should receive potentially distinct recommendations adapted to their individual tastes. Personalization is seldom attempted over catalogs of millions of items since the cost of existing personalization routines scale linearly in the number of candidate items. For example, performing two-sided personalized retrieval (with both event and item embeddings personalized to the user) incurs prohibitive storage and compute costs. Instead, it is common to use non-personalized retrieval to obtain a small shortlist of items over which personalized re-ranking can be done quickly. Despite being scalable, this strategy risks losing items uniquely relevant to a user that fail to get shortlisted during non-personalized retrieval. This paper bridges this gap by developing the XPERT algorithm that identifies a form of two-sided personalization that can be scalably implemented over millions of items and hundreds of millions of users. Key to overcoming the computational challenges of personalized retrieval is a novel concept of morph operators that can be used with arbitrary encoder architectures, completely avoids the steep memory overheads of two-sided personalization, provides millisecond-time inference and offers multi-intent retrieval. On multiple public and proprietary datasets, XPERT offered upto 5% superior recall and AUC than state-of-the-art techniques. Code for XPERT is available at https://github.com/personalizedretrieval/xpert.

Extreme Classification (XC) seeks to tag data points with the most relevant subset of labels from an extremely large label set. Performing deep XC with dense, learnt representations for data points and labels has attracted much attention due to its superiority over earlier XC methods that used sparse, hand-crafted features. Negative mining techniques have emerged as a critical component of all deep XC methods, allowing them to scale to millions of labels. However, despite recent advances, training deep XC models with large encoder architectures such as transformers remains challenging. This paper notices that memory overheads of popular negative mining techniques often force mini-batch sizes to remain small and slow training down. In response, this paper introduces NGAME, a light-weight mini-batch creation technique that offers provably accurate in-batch negative samples. This allows training with larger mini-batches offering significantly faster convergence and higher accuracies than existing negative sampling techniques. NGAME was found to be up to 16% more accurate than state-of-the-art methods on a wide array of benchmark datasets for extreme classification, as well as 3% more accurate at retrieving search engine queries in response to a user webpage visit to show personalized ads. In live A/B tests on a popular search engine, NGAME yielded up to 23% gains in click-through-rates. Code for NGAME is available at https://github.com/Extreme-classification/ngame

Scalability and accuracy are well recognized challenges in deep extreme multi-label learning where the objective is to train architectures for automatically annotating a data point with the most relevant subset of labels from an extremely large label set. This paper develops the DeepXML framework that addresses these challenges by decomposing the deep extreme multi-label task into four simpler sub-tasks each of which can be trained accurately and efficiently. Choosing different components for the four sub-tasks allows DeepXML to generate a family of algorithms with varying trade-offs between accuracy and scalability. In particular, DeepXML yields the Astec algorithm that could be 2-12% more accurate and 5-30x faster to train than leading deep extreme classifiers on publically available short text datasets. Astec could also efficiently train on Bing short text datasets containing up to 62 million labels while making predictions for billions of users and data points per day on commodity hardware. This allowed Astec to be deployed on the Bing search engine for a number of short text applications ranging from matching user queries to advertiser bid phrases to showing personalized ads where it yielded significant gains in click-through-rates, coverage, revenue and other online metrics over state-of-the-art techniques currently in production. DeepXML's code is available at https://github.com/Extreme-classification/deepxml.

Over the past decade, Determinantal Point Processes (DPPs) have proven to be a mathematically elegant framework for modeling diversity. Given a set of items N, DPPs define a probability distribution over subsets of N, with sets of larger diversity having greater probability. Recently, DPPs have achieved success in the domain of recommendation systems, as a method to enforce diversity of recommendations in addition to relevance. In large-scale recommendation applications however, the input typically comes in the form of a stream too large to fit into main memory. However, the natural greedy algorithm for DPP-based recommendations is memory intensive, and cannot be used in a streaming setting.

Recent studies have shown that both accuracy and explainability are important for recommendation. In this paper, we introduce explainable conversational recommendation, which enables incremental improvement of both recommendation accuracy and explanation quality through multi-turn user-model conversation. We show how the problem can be formulated, and design an incremental multi-task learning framework that enables tight collaboration between recommendation prediction, explanation generation, and user feedback integration. We also propose a multi-view feedback integration method to enable effective incremental model update. Empirical results demonstrate that our model not only consistently improves the recommendation accuracy but also generates explanations that fit user interests reflected in the feedbacks.

Existing algorithms for dictionary learning assume that the entries of the (high-dimensional) input data are fully observed. However, in several practical applications, only an incomplete fraction of the data entries may be available. For incomplete settings, no provably correct and polynomial-time algorithm has been reported in the dictionary learning literature. In this paper, we provide provable approaches for learning – from incomplete samples – a family of dictionaries whose atoms have sufficiently “spread-out” mass. First, we propose a descent-style iterative algorithm that linearly converges to the true dictionary when provided a sufficiently coarse initial estimate. Second, we propose an initialization algorithm that utilizes a small number of extra fully observed samples to produce such a coarse initial estimate. Finally, we theoretically analyze their performance and provide asymptotic statistical and computational guarantees.

Many applications of single channel source separation (SCSS) including automatic speech recognition (ASR), hearing aids etc. require an estimation of only one source from a mixture of many sources.Treating this special case as a regular SCSS problem where in all constituent sources are given equal priority in terms of reconstruction may result in a suboptimal separation performance.In this paper, we tackle the one source separation problem by suitably modifying the orthodox SCSS framework and focus only on one source at a time.The proposed approach is a generic framework that can be applied to any existing SCSS algorithm, improves performance, and scales well when there are more than two sources in the mixture unlike most existing SCSS methods.Additionally, existing SCSS algorithms rely on fine hyper-parameter tuning hence making them difficult to use in practice.Our framework takes a step towards automatic tuning of the hyper-parameters thereby making our method better suited for the mixture to be separated and thus practically more useful.We test our framework on a neural network based algorithm and the results show an improved performance in terms of SDR and SAR.

Inductive Matrix Completion (IMC) is an important class of matrix completion problems that allows direct inclusion of available features to enhance estimation capabilities. These models have found applications in personalized recommendation systems, multilabel learning, dictionary learning, etc. This paper examines a general class of noisy matrix completion tasks where the underlying matrix is following an IMC model i.e., it is formed by a mixing matrix (a priori unknown) sandwiched between two known feature matrices. The mixing matrix here is assumed to be well approximated by the product of two sparse matrices - referred here to as “sparse factor models.” We leverage the main theorem of [1] and extend it to provide theoretical error bounds for the sparsity-regularized maximum likelihood estimators for the class of problems discussed in this paper. The main result is general in the sense that it can be used to derive error bounds for various noise models. In this paper, we instantiate our main result for the case of Gaussian noise and provide corresponding error bounds in terms of squared loss.

We study the online constrained ranking problem motivated by an application to web-traffic shaping: an online stream of sessions arrive in which, within each session, we are asked to rank items. The challenge involves optimizing the ranking in each session so that local vs. global objectives are controlled: within each session one wishes to maximize a reward (local) while satisfying certain constraints over the entire set of sessions (global). A typical application of this setup is that of page optimization in a web portal. We wish to rank items so that not only is user engagement maximized in each session, but also other business constraints (such as the number of views/clicks delivered to various publishing partners) are satisfied.

This paper examines a general class of noisy matrix completion tasks, where the goal is to estimate a matrix from observations obtained at a subset of its entries, each of which is subject to random noise or corruption. Our specific focus is on settings where the matrix to be estimated is well-approximated by a product of two (a priori unknown) matrices, one of which is sparse. Such structural models-referred to here as sparse factor models-have been widely used, for example, in subspace clustering applications, as well as in contemporary sparse modeling and dictionary learning tasks. Our main theoretical contributions are estimation error bounds for sparsity-regularized maximum likelihood estimators for the problems of this form, which are applicable to a number of different observation noise or corruption models. Several specific implications are examined, including scenarios where observations are corrupted by additive Gaussian noise or additive heavier-tailed (Laplace) noise, Poisson-distributed observations, and highly quantized (e.g., 1 b) observations. We also propose a simple algorithmic approach based on the alternating direction method of multipliers for these tasks, and provide experimental evidence to support our error analyses.

Many machine learning applications employ a multiclass classification stage that uses multiple binary linear classifiers as building blocks. Among these, commonly used strategies such as one-vs-one classification can require learning a large number of hyperplanes, even when the number of classes to be discriminated among is modest. Further, when the data being classified is inherently high-dimensional, the storage and computational complexity associated with the application of multiple linear classifiers can ignite critical resource management issues. This work describes a novel multiclass classification method based on efficient use of a single “recycled” linear classifier (or ReLiC), which addresses these storage and implementation complexity issues. The proposed approach amounts to constraining the entire collection of hyperplanes to be circularly-shifted versions of each other, enabling classification procedures that may be implemented with efficient operations, such as circular convolution (which can be efficiently computed using transform domain techniques), and simple sampling/thresholding operations. We show that the optimization task associated with our proposed approach can be formulated as a quadratic program, and we introduce an efficient distributed procedure for its solution based on an alternating direction method of multipliers. Simulation results demonstrate that the performance of the proposed approach is comparable with the more complex, traditional multiclass linear classification strategies, suggesting the proposed approach is a viable alternative in large-scale data classification tasks.

Poisson processes are commonly used models for describing discrete arrival phenomena arising, for example, in photon-limited scenarios in low-light and infrared imaging, astronomy, and nuclear medicine applications. In this context, several recent efforts have evaluated Poisson denoising methods that utilize contemporary sparse modeling and dictionary learning techniques designed to exploit and leverage (local) shared structure in the images being estimated. This paper establishes a theoretical foundation for such procedures. Specifically, we formulate sparse and structured dictionary-based Poisson denoising methods as constrained maximum likelihood estimation strategies, and establish performance bounds for their mean-square estimation error using the framework of complexity penalized maximum likelihood analyses.

Recent breakthrough results in compressive sensing (CS) have established that many high dimensional signals can be accurately recovered from a relatively small number of non-adaptive linear observations, provided that the signals possess a sparse representation in some basis. Subsequent efforts have shown that the performance of CS can be improved by exploiting additional structure in the locations of the nonzero signal coefficients during inference or by utilizing some form of data-dependent adaptive measurement focusing during the sensing process. To the best of our knowledge, our own previous work was the first to establish the potential benefits that can be achieved when fusing the notions of adaptive sensing and structured sparsity. In that work, we examined the task of support recovery from noisy linear measurements, and established that an adaptive sensing strategy specifically tailored to signals that are tree-sparse can significantly outperform adaptive and non-adaptive sensing strategies that are agnostic to the underlying structure. In this paper, we establish fundamental performance limits for the task of support recovery of tree-sparse signals from noisy measurements, in settings where measurements may be obtained either non-adaptively (using a randomized Gaussian measurement strategy motivated by initial CS investigations) or by any adaptive sensing strategy. Our main results here imply that the adaptive tree sensing procedure analyzed in our previous work is nearly optimal, in the sense that no other sensing and estimation strategy can perform fundamentally better for identifying the support of tree-sparse signals.

Breakthrough results in compressive sensing (CS) have shown that high dimensional signals (vectors) can often be accurately recovered from a relatively small number of non-adaptive linear projection observations, provided that they possess a sparse representation in some basis. Subsequent efforts have established that the reconstruction performance of CS can be improved by employing additional prior signal knowledge, such as dependency in the location of the non-zero signal coefficients (structured sparsity) or by collecting measurements sequentially and adaptively, in order to focus measurements into the proper subspace where the unknown signal resides. In this paper, we examine a powerful hybrid of adaptivity and structure. We identify a particular form of structured sparsity that is amenable to adaptive sensing, and using concepts from sparse hierarchical dictionary learning we demonstrate that sparsifying dictionaries exhibiting the appropriate form of structured sparsity can be learned from a collection of training data. The combination of these techniques (structured dictionary learning and adaptive sensing) results in an effective and efficient adaptive compressive acquisition approach which we refer to as LASeR (Learning Adaptive Sensing Representations).

Estimating the level set of a signal from measurements is a task that arises in a variety of fields, including medical imaging, astronomy, and digital elevation mapping. Motivated by scenarios where accurate and complete measurements of the signal may not available, we examine here a simple procedure for estimating the level set of a signal from highly incomplete measurements, which may additionally be corrupted by additive noise. The proposed procedure is based on box-constrained Total Variation (TV) regularization. We demonstrate the performance of our approach, relative to existing state-of-the-art techniques for level set estimation from compressive measurements, via several simulation examples.