In this week's edition: Elon Musk suffers the slings and arrows of outrageous fortune, Apple beats itself up over Siri, and Meta goes after one of its own over a tell-all book. The past 10 days have marked several of the most significant setbacks for Musk in months. Tesla, arguably his marquee company, continued to fall in value as investors worried about the threat of trade war and possible recession – as well as declining profits. Escalating protests against the company over the billionaire's role in the government also grew in number and intensity across the US, coupled with rising cases of vandalism and social stigma against his cars. SpaceX has also struggled, with one of its rockets dramatically exploding in midflight last week and then an announcement that it was delaying a rescue mission to retrieve "stranded" astronauts. The company tried again two days later.

Deep models in industrial applications rely on thousands of features for accurate predictions, such as deep recommendation systems. While new features are introduced to capture evolving user behavior, outdated or redundant features often remain, significantly increasing storage and computational costs. To address this issue, feature selection methods are widely adopted to identify and remove less important features. However, existing approaches face two major challenges: (1) they often require complex hyperparameter (Hp) tuning, making them difficult to employ in practice, and (2) they fail to produce well-separated feature importance scores, which complicates straightforward feature removal. Moreover, the impact of removing unimportant features can only be evaluated through retraining the model, a time-consuming and resource-intensive process that severely hinders efficient feature selection. To solve these challenges, we propose a novel feature selection approach, ShuffleGate. In particular, it shuffles all feature values across instances simultaneously and uses a gating mechanism that allows the model to dynamically learn the weights for combining the original and shuffled inputs. Notably, it can generate well-separated feature importance scores and estimate the performance without retraining the model, while introducing only a single Hp. Experiments on four public datasets show that our approach outperforms state-of-the-art methods in feature selection for model retraining. Moreover, it has been successfully integrated into the daily iteration of Bilibili's search models across various scenarios, where it significantly reduces feature set size (up to 60%+) and computational resource usage (up to 20%+), while maintaining comparable performance.

In modern social media, recommender systems (RecSys) rely on the click-through rate (CTR) as the standard metric to evaluate user engagement. CTR prediction is traditionally framed as a binary classification task to predict whether a user will interact with a given item. However, this approach overlooks the complexity of real-world social modeling, where the user, item, and their interactive features change dynamically in fast-paced online environments. This dynamic nature often leads to model instability, reflected in overfitting short-term fluctuations rather than higher-level interactive patterns. While overfitting calls for more scaled and refined supervisions, current solutions often rely on binary labels that overly simplify fine-grained user preferences through the thresholding process, which significantly reduces the richness of the supervision. Therefore, we aim to alleviate the overfitting problem by increasing the supervision bandwidth in CTR training. Specifically, (i) theoretically, we formulate the impact of fine-grained preferences on model stability as a Lipschitz constrain; (ii) empirically, we discover that scaling the supervision bandwidth can act as an implicit Lipschitz regularizer, stably optimizing existing CTR models to achieve better generalizability. Extensive experiments show that this scaled supervision significantly and consistently improves the optimization process and the performance of existing CTR models, even without the need for additional hyperparameter tuning.

Graph Neural Networks have significantly advanced research in recommender systems over the past few years. These methods typically capture global interests using aggregated past interactions and rely on static embeddings of users and items over extended periods of time. While effective in some domains, these methods fall short in many real-world scenarios, especially in finance, where user interests and item popularity evolve rapidly over time. To address these challenges, we introduce a novel extension to Light Graph Convolutional Network (LightGCN) designed to learn temporal node embeddings that capture dynamic interests. Our approach employs causal convolution to maintain a forward-looking model architecture. By preserving the chronological order of user-item interactions and introducing a dynamic update mechanism for embeddings through a sliding window, the proposed model generates well-timed and contextually relevant recommendations. Extensive experiments on a real-world dataset from BNP Paribas demonstrate that our approach significantly enhances the performance of LightGCN while maintaining the simplicity and efficiency of its architecture. Our findings provide new insights into designing graph-based recommender systems in time-sensitive applications, particularly for financial product recommendations.

Amazon's Alexa service is rolling out on March 28, and with it supposedly comes a more personalized, intuitive, and powerful digital assistant thanks to its underlying generative AI technology. But for the new features to work, the company is asking a lot from its Echo and smart device users--whether or not they choose to use Alexa at all. Alexa is billed as a major upgrade that includes individual voice recognition through Alexa Voice ID, nuanced calendar scheduling, Ring home security system integrations, and product purchasing capabilities. It's Amazon's latest effort to generate a profit from Alexa, which lost 25 billion in revenue between 2007-2021 according to The Wall Street Journal last year. While Alexa will be added to all Prime subscriptions, users without Prime can enroll in the program for 19.99 per month.

Now that Apple's recent slew of hardware releases are behind us, we got some news on the software side last week. First, the company publicly announced that it was delaying the smarter, more personal version of Siri that'll be powered by Apple Intelligence. Then, rumors sprang up again that Apple was giving an extensive visual update to its software platforms, including iOS 19 and macOS 16 which are expected to be revealed at WWDC in June. The sources for this redesign rumor are solid. Jon Prosser dropped a video on his YouTube channel Front Page Tech back in January where he said that he had seen a redesigned Camera app for the next version of iOS that had a number of interface changes that made it feel more like a visionOS app. His thinking is that Apple wouldn't redesign a core app like Camera without bringing changes to some of the rest of the OS, as well.

Time is running out for Google Assistant as Gemini prepares to take its place on mobile and--eventually--smart devices. Now Google is announcing another round of features that Google Assistant is soon to lose. None of the about-to-be-yanked features are all that critical, but the move is yet another sign that Google Assistant is going by the wayside. The nixed features were spotted by 9to5Google on a support page that lists other deprecated Google Assistant features, including more than a dozen that were dropped early last year. Among the chopped Google Assistant features that owners of Nest smart speakers and displays might miss is Family Bell, which allowed users to create reminder bells for family events such as breakfast or dinner time.

But if you have an Amazon Echo, there's bad news for you - as Amazon is about to controversially kill a popular privacy feature. Until now, some Amazon Echo devices have had the option to process commands locally'on-device', keeping your voice within the confines of your home. But from March 28, all Alexa-powered Echo smart speakers will send your voice recordings to the cloud, whether you like it or not. Cory Doctorow, a blogger and expert on digital rights management, called it'absolutely unforgivable' because it will let Amazon workers snoop on all Echo recordings. Amazon has already received criticism for storing conversations users have with Alexa, which have been listened to and transcribed by staff, it admitted in 2019.

The rapid advancement of artificial intelligence (AI) technologies has led to an increasing deployment of AI models on edge and terminal devices, driven by the proliferation of the Internet of Things (IoT) and the need for real-time data processing. This survey comprehensively explores the current state, technical challenges, and future trends of on-device AI models. We define on-device AI models as those designed to perform local data processing and inference, emphasizing their characteristics such as real-time performance, resource constraints, and enhanced data privacy. The survey is structured around key themes, including the fundamental concepts of AI models, application scenarios across various domains, and the technical challenges faced in edge environments. We also discuss optimization and implementation strategies, such as data preprocessing, model compression, and hardware acceleration, which are essential for effective deployment. Furthermore, we examine the impact of emerging technologies, including edge computing and foundation models, on the evolution of on-device AI models. By providing a structured overview of the challenges, solutions, and future directions, this survey aims to facilitate further research and application of on-device AI, ultimately contributing to the advancement of intelligent systems in everyday life.

Matrix completion is fundamental for predicting missing data with a wide range of applications in personalized healthcare, e-commerce, recommendation systems, and social network analysis. Traditional matrix completion approaches typically assume centralized data storage, which raises challenges in terms of computational efficiency, scalability, and user privacy. In this paper, we address the problem of federated matrix completion, focusing on scenarios where user-specific data is distributed across multiple clients, and privacy constraints are uncompromising. Federated learning provides a promising framework to address these challenges by enabling collaborative learning across distributed datasets without sharing raw data. We propose \texttt{FedMC-ADMM} for solving federated matrix completion problems, a novel algorithmic framework that combines the Alternating Direction Method of Multipliers with a randomized block-coordinate strategy and alternating proximal gradient steps. Unlike existing federated approaches, \texttt{FedMC-ADMM} effectively handles multi-block nonconvex and nonsmooth optimization problems, allowing efficient computation while preserving user privacy. We analyze the theoretical properties of our algorithm, demonstrating subsequential convergence and establishing a convergence rate of $\mathcal{O}(K^{-1/2})$, leading to a communication complexity of $\mathcal{O}(\epsilon^{-2})$ for reaching an $\epsilon$-stationary point. This work is the first to establish these theoretical guarantees for federated matrix completion in the presence of multi-block variables. To validate our approach, we conduct extensive experiments on real-world datasets, including MovieLens 1M, 10M, and Netflix. The results demonstrate that \texttt{FedMC-ADMM} outperforms existing methods in terms of convergence speed and testing accuracy.