Cloud Workload Prediction (CWP) is a critical task in cloud computing, essential for resource scheduling, performance optimization, and cost management. However, existing time series prediction methods struggle with instability and inefficiency when applied directly to cloud workloads due to their high variability and frequent fluctuations. To address these challenges, we propose DEL4CW, a novel D eep E xpansion L earning framework specifically designed for CWP . DEL4CW introduces a unique self-decoupling mechanism to disentangle the complex dependencies present in highly variable cloud workloads, leading to more accurate predictions of job arrival rates. The core contribution of DEL4CW lies in its ability to decouple cloud workload signals into three key components—trend, periodicity, and residuals—by treating these as hidden variables. This enables the model to better manage both short-term fluctuations and long-term workload trends. DEL4CW employs a deep expansion learning framework structured as stacked blocks, where each block includes dedicated modules for trend, periodicity, and compensation. Specifically, the trend module utilizes multi-layer fully connected networks to capture evolving trends at multiple granularities, while the periodicity module leverages multi-head attention to identify diverse periodic patterns. The compensation module addresses unpredictable, localized fluctuations, improving the model’s robustness to noise. In addition to its predictive accuracy, DEL4CW provides interpretable insights through its hierarchical design, allowing for layer-by-layer aggregation of meaningful partial predictions. This interpretability stems from the doubly residual learning pipeline, which ensures that each prediction block contributes progressively refined predictions. Extensive experiments on real-world cloud workload traces demonstrate that DEL4CW significantly outperforms existing baselines, with error reductions reaching up to 27.74% in certain scenarios.

As recommender systems are essential to various web domains such as e-commerce and web content sharing, providing equitable item exposure regardless of popularity becomes an imperative requirement. However, traditional fairness-aware approaches typically aim to achieve a better tradeoff between recommendation accuracy and fairness, and focus on improving the exposure rate of the long-tail items on static settings, evaluating fairness on one-shot recommendation decisions using logged data. Such methods overlook the dynamic nature of user preferences in real-world interactive environments. In contrast, our work seeks a win-win solution that simultaneously enhances recommendation accuracy and fairness over the long term, rather than merely trading off one against the other. To achieve this goal, we empirically demonstrate and analyze the spatiotemporal heterogeneity of user popularity preference. Our findings reveal complementary characteristics that, when fully exploited, can guide personalized strategies for long-term fairness. Building on this insight, we propose HER4IF, a novel hierarchical reinforcement learning framework designed for interactive recommendation. HER4IF decomposes the recommendation process into two key tasks: dynamic fairness control and item recommendation. The high-level agent continuously learns adaptive fairness constraints from evolving user popularity preferences, while the low-level agent refines recommendation policies under these personalized constraints. Extensive experiments on three real-world datasets and the interactive recommendation platform KuaiSim demonstrate that HER4IF significantly outperforms state-of-the-art methods, achieving substantial improvements in both fairness and recommendation accuracy. Our code is available at: https://github.com/1163710212/HER4IF .

It is a longstanding problem that Q-learning suffers from the overestimation bias. This issue originates from the fact that Q-learning uses the expectation of maximum Q-value to approximate the maximum expected Q-value. A number of algorithms, such as Double Q-learning, were proposed to address this problem by reducing the estimation of maximum Q-value, but this may lead to an underestimation bias. Note that this underestimation bias may have a larger performance penalty than the overestimation bias. Different from previous algorithms, this article studies this issue from a fresh perspective, i.e., meta-learning view, which leads to our Meta-Debias Q-learning. The main idea is to extract the maximum expected Q-value with meta-learning over multiple tasks to remove the estimation bias of maximum Q-value and help the agent choose the optimal action more accurately. However, there are two challenges: (1) How to automatically select suitable training tasks? (2) How to positively transfer the meta-knowledge from selected tasks to remove the estimation bias of maximum Q-value? To address the two challenges mentioned above, we quantify the similarity between the training tasks and the test task. This similarity enables us to select appropriate “partial” training tasks and helps the agent extract the maximum expected Q-value to remove the estimation bias. Extensive experiment results show that our Meta-Debias Q-learning outperforms SOTA baselines drastically in three evaluation indicators, i.e., maximum Q-value, policy, and reward. More specifically, our Meta-Debias Q-learning only underestimates \(1.2*10^{-3}\) than the maximum expected Q-value in the multi-armed bandit environment and only differs \(5.04\%-5\%=0.04\%\) than the optimal policy in the two states MDP environment. In addition, we compare the uniform weight and our similarity weight. Experiment results reveal fundamental insights into why our proposed algorithm outperforms in the maximum Q-value, policy, and reward.

Online rating system serves as an indispensable building block for many web applications. Previous studies showed that due to assimilate-contrast effects, historical ratings could significantly distort users' ratings, leading to low accuracy of product quality estimation and recommendation. To understand assimilate-contrast effects, an “accurate” model is still missing as previous models do not capture important factors like rating recency, selection bias, etc. Furthermore, an analytical framework to characterize product estimation accuracy under assimilate-contrast effects is also missing. This paper aims to fill in this gap. We propose a probabilistic model to quantify the aforementioned important factors on assimilate-contrast effects. We apply stochastic approximation theory to show that when the rating bias satisfies mild contraction conditions, the aggregate rating converges under aggregate opinion heterogeneity. We also apply non-stationary Markov chain theory to show that when the strength of assimilate-contrast satisfies mild stable conditions, the aggregate rating converges under rating recency. We also derive an equation to characterize the converged aggregate ratings. These conditions reveal important insights on how the aforementioned factors influence the convergence and guide the online rating system operator to design appropriate rating aggregation rules and rating displaying strategies. We apply it to rating prediction tasks and product recommendation tasks. Experiment results on four public datasets show that our model can improve the rating prediction and recommendation accuracy over previous models significantly, under various metrics like RMSE, NDCG, etc. We also demonstrate the flexibility of our model by showing that it can be applied to enhance other rating behavior models.

While personalization increases the utility of item recommendation, it also suffers from the issue of popularity bias. However, previous methods emphasize adopting supervised learning models to relieve popularity bias in the static recommendation, ignoring the dynamic transfer of user preference and amplification effects of the feedback loop in the recommender system (RS). In this paper, we focus on studying this issue in the interactive recommendation. We argue that diversification and novelty are both equally crucial for improving user satisfaction of IRS in the aforementioned setting. To achieve this goal, we propose a D iversity- N ovelty- a ware I nteractive R ecommendation framework (DNaIR) that augments offline reinforcement learning (RL) to increase the exposure rate of long-tail items with high quality. Its main idea is first to aggregate the item similarity, popularity, and quality into the reward model to help the planning of RL policy. It then designs a diversity-aware stochastic action generator to achieve an efficient and lightweight DNaIR algorithm. Extensive experiments are conducted on the three real-world datasets and an authentic RL environment (Virtual-Taobao). The experiments show that our model can better and full use of the long-tail items to improve recommendation satisfaction, especially those low popularity items with high-quality ones, thus achieving state-of-the-art performance.

A high-dimensional and incomplete (HDI) matrix is a common form of big data in most industrial applications. Stochastic gradient descent (SGD) algorithm optimized latent factor analysis (LFA) model is often adopted in learning the abundant knowledge in HDI matrix. Despite its computational tractability and scalability, when solving a bilinear problem such as LFA, the regular SGD algorithm tends to be stuck in a local optimum. To address this issue, the paper innovatively proposes an Adjusted Stochastic Gradient Descent (ASGD) for Latent Factor Analysis, where the adjustment mechanism is implemented by considering the bi-polar gradient directions during optimization, such mechanism is theoretically proved for its efficiency in overstepping local saddle points and avoiding premature convergence. Also, the hyper-parameters of the model are implemented in a self-adaptive manner using the particle swarm optimization (PSO) algorithm, for higher practicality. Experimental results show that the proposed model outperforms other state-of-the-art approaches on six different HDI matrices from industrial applications, especially in prediction accuracy for missing data.

How to accurately predict unknown quality-of-service (QoS) data based on observed ones is a hot yet thorny issue in Web service-related applications. Recently, a latent factor (LF) model has shown its efficiency in addressing this issue owing to its high accuracy and scalability. An LF model can be improved by identifying user and service neighborhoods based on user and service geographical information. However, such information can be difficult to acquire in most applications with the considerations of information security, identity privacy, and commercial interests in a real system. Besides, the existing LF model-based QoS predictors mostly ignore the reliability of given QoS data where noises commonly exist to cause accuracy loss. To address the above issues, this paper proposes a data-characteristic-aware latent factor (DCALF) model to implement highly accurate QoS predictions, where ‘data-characteristic-aware’ indicates that it can appropriately implement QoS prediction according to the characteristics of given QoS data. Its main idea is two-fold: a) it detects the neighborhoods and noises of users and services based on the dense LFs extracted from the original sparse QoS data, b) it incorporates a density peaks-based clustering method into its modeling process for achieving the simultaneous detections of both neighborhoods and noises of QoS data. With such designs, it precisely represents the given QoS data in spite of their sparsity, thereby achieving highly accurate predictions for unknown ones. Experimental results on two QoS datasets generated by real-world Web services demonstrate that the proposed DCALF model outperforms state-of-the-art QoS predictors, making it highly competitive in addressing the issue of Web service selection and recommendation.

Recommender systems (RSs) commonly describe its user-item preferences with a high-dimensional and sparse (HiDS) matrix filled with non-negative data. A non-negative latent factor (NLF) model relying on a single latent factor-dependent, non-negative and multiplicative update (SLF-NMU) algorithm is frequently adopted to process such an HiDS matrix. However, an NLF model mostly adopts Euclidean distance for its objective function, which is naturally a special case of α-β-divergence. Moreover, it frequently suffers slow convergence. For addressing these issues, this study proposes a generalized and fast-converging non-negative latent factor (GFNLF) model. Its main idea is two-fold: a) adopting α-β-divergence for its objective function, thereby enhancing its representation ability for HiDS data; b) deducing its momentum-incorporated non-negative multiplicative update (MNMU) algorithm, thereby achieving its fast convergence. Empirical studies on two HiDS matrices emerging from real RSs demonstrate that with carefully-tuned hyperparameters, a GFNLF model outperforms state-of-the-art models in both computational efficiency and prediction accuracy for missing data of an HiDS matrix.

Online streaming feature selection has received extensive attention in the past few years. Existing approaches have a common assumption that the feature space of the fixed data instances increases dynamically without any missing entry. This assumption, however, does not always hold in many real-world applications. For example, in a credit evaluation system, we cannot collect the complete dynamic features for each person and/or enterprise. Motivated by this observation, this paper aims at conducting online feature selection from capricious streaming features, where features flow in one by one with some random missing entries while the number of data instances remains fixed. To do so, we propose a general framework named GF-CSF. The main idea of GF-CSF is to adopt latent factor analysis to preprocess capricious streaming features for completing their missing entries before conducting feature selection. Both theoretical and experimental analyses indicate that GF-CSF can efficiently improve any existing model of online streaming features selection to achieve online capricious streaming features selection.

High-dimensional and sparse (HiDS) matrices are commonly encountered in many big data-related industrial applications like recommender systems. When acquiring useful patterns from them, non-negative matrix factorization (NMF) models have proven to be highly effective because of their fine representativeness of non-negative data. However, current NMF techniques suffer from a) inefficiency in addressing HiDS matrices, and b) constrained training schemes lack of flexibility, extensibility and adaptability. To address these issues, this work proposes to factorize industrial-size sparse matrices via a novel Inherently Non-negative Latent Factor (INLF) model. It connects the output factors and decision variables via a single-element-dependent sigmoid function, thereby innovatively removing the non-negativity constraints from its training process without impacting the solution accuracy. Hence, its training process is unconstrained, highly flexible and compatible with general learning schemes. Experimental results on five HiDS matrices generated by industrial applications indicate that INLF is able to acquire non-negative latent factors from them in a more efficient manner than any existing method does.