This paper analyzes Cross-Entropy (CE) loss in knowledge distillation (KD) for recommender systems. KD for recommender systems targets at distilling rankings, especially among items most likely to be preferred, and can only be computed on a small subset of items. Considering these features, we reveal the connection between CE loss and NDCG in the field of KD. We prove that when performing KD on an item subset, minimizing CE loss maximizes the lower bound of NDCG, only if an assumption of closure is satisfied. It requires that the item subset consists of the student's top items. However, this contradicts our goal of distilling rankings of the teacher's top items. We empirically demonstrate the vast gap between these two kinds of top items. To bridge the gap between our goal and theoretical support, we propose Rejuvenated Cross-Entropy for Knowledge Distillation (RCE-KD). It splits the top items given by the teacher into two subsets based on whether they are highly ranked by the student. For the subset that defies the condition, a sampling strategy is devised to use teacher-student collaboration to approximate our assumption of closure. We also combine the losses on the two subsets adaptively. Extensive experiments demonstrate the effectiveness of our method. Our code is available at https://anonymous.4open.science/r/RCE-KD.

When you purchase through links in our articles, we may earn a small commission. Perplexity Email Assistant is like a virtual assistant for your inbox. AI company Perplexity, best known for its free AI-powered answer engine, is now launching another AI-powered tool called Perplexity Email Assistant . This AI assistant integrates directly into your email inbox and can help you maintain better control over your email. Email Assistant can write email drafts in your own tone and conversational style, organize messages, suggest meeting times, and even participate in email threads to save you time on long exchanges.

'ChatGPT, what stocks should I buy?' AI fuels boom in robo-advisory market Stock picking using ChatGPT requires some financial knowledge and even its adopters say there is a high risk of getting it wrong before getting it right. LONDON - As ChatGPT nears its third birthday, at least one in 10 retail investors is using a chatbot to pick stocks, fueling a boom in the robo-advisory market, but even fans say it is a high-risk strategy that cannot replace traditional advisers just yet. Thanks to artificial intelligence, anyone can select stocks, monitor them and obtain investment analysis that was once only available to big banks or institutional investors. The robo-advisory market -- which includes all companies providing automated, algorithm-driven financial advice such as fintech, banks and wealth managers -- is forecast to grow to $470.91 billion in revenues in 2029 from $61.75 billion last year, marking a roughly 600% increase, according to data analysis firm Research and Markets. In a time of both misinformation and too much information, quality journalism is more crucial than ever.

'People say I come across as incredibly boring!' How to find love on the dating apps - whatever the obstacles Sick of swiping and messaging but never meeting anyone you like and who likes you back? Here's what worked for some lucky couples U sing dating apps to find love is commonplace these days - and yet, for many singles, it has become a double-edged sword. The perks of having a never-ending supply of potential matches at your fingertips are obvious - but the appeal of connecting and meeting with strangers is time-limited. It can be especially frustrating to feel as if you're stuck at the swiping stage. In 2023, US jeweller Shane Company found that the average American will spend about eight months using dating apps - swiping on around 3,960 profiles - before finding a partner.

Multimodal data has significantly advanced recommendation systems by integrating diverse information sources to model user preferences and item characteristics. However, these systems often struggle with redundant and irrelevant information, which can degrade performance. Most existing methods either fuse multimodal information directly or use rigid architectural separation for disentanglement, failing to adequately filter noise and model the complex interplay between modalities. To address these challenges, we propose a novel framework, the Multimodal Representation-disentangled Information Bottleneck (MRdIB). Concretely, we first employ a Multimodal Information Bottleneck to compress the input representations, effectively filtering out task-irrelevant noise while preserving rich semantic information. Then, we decompose the information based on its relationship with the recommendation target into unique, redundant, and synergistic components. We achieve this decomposition with a series of constraints: a unique information learning objective to preserve modality-unique signals, a redundant information learning objective to minimize overlap, and a synergistic information learning objective to capture emergent information. By optimizing these objectives, MRdIB guides a model to learn more powerful and disentangled representations. Extensive experiments on several competitive models and three benchmark datasets demonstrate the effectiveness and versatility of our MRdIB in enhancing multimodal recommendation.

The "No Free Lunch" theorem dictates that no single recommender algorithm is optimal for all users, creating a significant Algorithm Selection Problem. Standard meta-learning approaches aim to solve this by selecting an algorithm based on user features, but treat the fundamentally diverse algorithms themselves as equivalent, "black-box" choices. This thesis investigates the impact of overcoming this limitation by engineering a comprehensive feature set to explicitly characterize the algorithms themselves. We combine static code metrics, Abstract Syntax Tree properties, behavioral performance landmarks, and high-level conceptual features. We evaluate two meta-learners across five datasets: a baseline using only user features and our proposed model using both user and algorithm features. Our results show that the meta-learner augmented with algorithm features achieves an average NDCG@10 of 0.143, a statistically significant improvement of 11.7% over the Single Best Algorithm baseline (0.128). However, we found that the inclusion of algorithm features did not lead to an improvement in overall NDCG@10 over the meta learner using only user features (0.144). While adding algorithm features to the meta-learner did improve its Top-1 selection accuracy (+16.1%), this was counterbalanced by leading to a lower Top-3 accuracy (-10.7%). We conclude that for the per-user algorithm selection task in recommender systems, the predictive power of user features is overwhelmingly dominant. While algorithm features improve selection precision, unlocking their potential to boost overall performance remains a non-trivial challenge.

Graph-based recommendation systems are effective at modeling collaborative patterns but often suffer from two limitations: overreliance on low-pass filtering, which suppresses user-specific signals, and omission of sequential dynamics in graph construction. We introduce GSPRec, a graph spectral model that integrates temporal transitions through sequentially-informed graph construction and applies frequency-aware filtering in the spectral domain. GSPRec encodes item transitions via multi-hop diffusion to enable the use of symmetric Laplacians for spectral processing. To capture user preferences, we design a dual-filtering mechanism: a Gaussian bandpass filter to extract mid-frequency, user-level patterns, and a low-pass filter to retain global trends. Extensive experiments on four public datasets show that GSPRec consistently outperforms baselines, with an average improvement of 6.77% in NDCG@10. Ablation studies show the complementary benefits of both sequential graph augmentation and bandpass filtering.

The pre-ranking stage in industrial recommendation systems faces a fundamental conflict between efficiency and effectiveness. While powerful models like Target Attention (TA) excel at capturing complex feature interactions in the ranking stage, their high computational cost makes them infeasible for pre-ranking, which often relies on simplistic vector-product models. This disparity creates a significant performance bottleneck for the entire system. To bridge this gap, we propose TARQ, a novel pre-ranking framework. Inspired by generative models, TARQ's key innovation is to equip pre-ranking with an architecture approximate to TA by Residual Quantization. This allows us to bring the modeling power of TA into the latency-critical pre-ranking stage for the first time, establishing a new state-of-the-art trade-off between accuracy and efficiency. Extensive offline experiments and large-scale online A/B tests at Taobao demonstrate TARQ's significant improvements in ranking performance. Consequently, our model has been fully deployed in production, serving tens of millions of daily active users and yielding substantial business improvements.

When you purchase through links in our articles, we may earn a small commission. You can now control your Google Home-connected devices straight from a web browser, no app required. A truckload of exciting Google Home news is expected to hit next week, including new smart devices and more details about Gemini for Home. But Google still managed to sneak in a different smart home announcement week, and it's a welcome one. The news focuses on the web-based version of Google Home, which just added a major new feature: the ability to control your Google Home-connected devices directly over the web.

Abstract--Most educational recommender systems are tuned and judged on click-or rating-based relevance, leaving their true pedagogical impact unclear . We introduce OBER--an Outcome-Based Educational Recommender that embeds learning outcomes and assessment items directly into the data schema, so any algorithm can be evaluated on the mastery it fosters. OBER uses a minimalist entity-relation model, a log-driven mastery formula, and a plug-in architecture. Integrated into an e-learning system in non-formal domain, it was evaluated trough a two-week A/B/C test with over 5 700 learners across three methods: fixed expert trajectory, collaborative filtering (CF), and knowledge-based (KB) filtering. CF maximized retention, but the fixed path achieved the highest mastery. Because OBER derives business, relevance, and learning metrics from the same logs, it lets practitioners weigh relevance and engagement against outcome mastery with no extra testing overhead. The framework is method-agnostic and readily extensible to future adaptive or context-aware recommenders. Index T erms--recommendation systems, e-learning, evaluation, assessment, intended learning outcomes, constructive alingment, empirical software engineering.