www.interconnects.ai","siteUrl":"https://www.interconnects.ai/podcast","image":null,"errorMessage":null,"errorAt":null,"ownerUserId":null}},{"feedId":"41459996870678533","id":"134711480750062592","title":"Understanding Aggregate Trends for Apple Intelligence Using Differential Privacy","url":"https://machinelearning.apple.com/research/differential-privacy-aggregate-trends","content":"At Apple, we believe privacy is a fundamental human right. And we believe in giving our users a great experience while protecting their privacy. For years, we’ve used techniques like differential privacy as part of our opt-in device analytics program. This lets us gain insights into how our products are used, so we can improve them, while protecting user privacy by preventing Apple from seeing individual-level data from those users.\\nThis same need to understand usage while protecting privacy is also present in Apple Intelligence. One of our principles is that Apple does not use our users\'…","description":"At Apple, we believe privacy is a fundamental human right. And we believe in giving our users a great experience while protecting their privacy. For years, we’ve used techniques like differential privacy as part of our opt-in device analytics program. This lets us gain insights…","guid":"differential-privacy-aggregate-trends","author":null,"authorUrl":null,"authorAvatar":null,"insertedAt":"2025-04-14T20:58:54.824Z","publishedAt":"2025-04-14T00:00:00.992Z","media":null,"categories":null,"attachments":null,"extra":null,"language":null,"feeds":{"type":"feed","id":"41459996870678533","url":"https://machinelearning.apple.com/rss.xml","title":"Apple Machine Learning Research","description":"Apple machine learning teams are engaged in state of the art research in machine learning and artificial intelligence. Learn about the latest advancements.","siteUrl":"https://machinelearning.apple.com/","image":null,"errorMessage":null,"errorAt":null,"ownerUserId":null}},{"feedId":"41459996870678533","id":"135041670538433536","title":"FocalLens: Instruction Tuning Enables Zero-Shot Conditional Image Representations","url":"https://machinelearning.apple.com/research/focallens","content":"This paper was accepted at the Workshop on Foundation Models in the Wild at ICLR 2025.\\nVisual understanding is inherently contextual - what we focus on in an image depends on the task at hand. For instance, given an image of a person holding a bouquet of flowers, we may focus on either the person such as their clothing, or the type of flowers, depending on the context of interest. Yet, most existing image encoding paradigms represent an image as a fixed, generic feature vector, overlooking the potential needs of prioritizing varying visual information for different downstream use cases. In…","description":"This paper was accepted at the Workshop on Foundation Models in the Wild at ICLR 2025. Visual understanding is inherently contextual - what we focus on in an image depends on the task at hand. For instance, given an image of a person holding a bouquet of flowers, we may focus on…","guid":"focallens","author":null,"authorUrl":null,"authorAvatar":null,"insertedAt":"2025-04-15T18:50:58.205Z","publishedAt":"2025-04-14T00:00:00.656Z","media":null,"categories":null,"attachments":null,"extra":null,"language":null,"feeds":{"type":"feed","id":"41459996870678533","url":"https://machinelearning.apple.com/rss.xml","title":"Apple Machine Learning Research","description":"Apple machine learning teams are engaged in state of the art research in machine learning and artificial intelligence. Learn about the latest advancements.","siteUrl":"https://machinelearning.apple.com/","image":null,"errorMessage":null,"errorAt":null,"ownerUserId":null}},{"feedId":"41459996870678533","id":"134790670695109634","title":"Language Models Know More Than They Show: Exploring Hallucinations From the Model\'s Viewpoint","url":"https://machinelearning.apple.com/research/exploring-hallucinations","content":"Large language models (LLMs) often produce errors, including factual inaccuracies, biases, and reasoning failures, collectively referred to as \\"hallucinations\\". Recent studies have demonstrated that LLMs\' internal states encode information regarding the truthfulness of their outputs, and that this information can be utilized to detect errors. In this work, we show that the internal representations of LLMs encode much more information about truthfulness than previously recognized. We first discover that the truthfulness information is concentrated in specific tokens, and leveraging this…","description":"Large language models (LLMs) often produce errors, including factual inaccuracies, biases, and reasoning failures, collectively referred to as \\"hallucinations\\". Recent studies have demonstrated that LLMs\' internal states encode information regarding the truthfulness of their…","guid":"exploring-hallucinations","author":null,"authorUrl":null,"authorAvatar":null,"insertedAt":"2025-04-15T02:13:35.177Z","publishedAt":"2025-04-11T00:00:00.341Z","media":null,"categories":null,"attachments":null,"extra":null,"language":null,"feeds":{"type":"feed","id":"41459996870678533","url":"https://machinelearning.apple.com/rss.xml","title":"Apple Machine Learning Research","description":"Apple machine learning teams are engaged in state of the art research in machine learning and artificial intelligence. Learn about the latest advancements.","siteUrl":"https://machinelearning.apple.com/","image":null,"errorMessage":null,"errorAt":null,"ownerUserId":null}},{"feedId":"41459996870678533","id":"133315566608402432","title":"MM-Ego: Towards Building Egocentric Multimodal LLMs","url":"https://machinelearning.apple.com/research/mm-ego","content":"This research aims to comprehensively explore building a multimodal foundation model for egocentric video understanding. To achieve this goal, we work on three fronts. First, as there is a lack of QA data for egocentric video understanding, we automatically generate 7M high-quality QA samples for egocentric videos ranging from 30 seconds to one hour long in Ego4D based on human-annotated data. This is one of the largest egocentric QA datasets. Second, we contribute a challenging egocentric QA benchmark with 629 videos and 7,026 questions to evaluate the models\' ability in recognizing and…","description":"This research aims to comprehensively explore building a multimodal foundation model for egocentric video understanding. To achieve this goal, we work on three fronts. First, as there is a lack of QA data for egocentric video understanding, we automatically generate 7M high…","guid":"mm-ego","author":null,"authorUrl":null,"authorAvatar":null,"insertedAt":"2025-04-11T00:32:02.958Z","publishedAt":"2025-04-11T00:00:00.132Z","media":null,"categories":null,"attachments":null,"extra":null,"language":null,"feeds":{"type":"feed","id":"41459996870678533","url":"https://machinelearning.apple.com/rss.xml","title":"Apple Machine Learning Research","description":"Apple machine learning teams are engaged in state of the art research in machine learning and artificial intelligence. Learn about the latest advancements.","siteUrl":"https://machinelearning.apple.com/","image":null,"errorMessage":null,"errorAt":null,"ownerUserId":null}},{"feedId":"52351163392512007","id":"133169567957908480","title":"矩阵的有效秩(Effective Rank)","url":"https://kexue.fm/archives/10847","content":"秩(Rank)是线性代数中的重要概念,它代表了矩阵的内在维度。然而,数学上对秩的严格定义,很多时候并不完全适用于数值计算场景,因为秩等于非零奇异值的个数,而数学上对“等于零”这件事的理解跟数值计算有所不同,数学上的“等于零”是绝对地、严格地等于零,哪怕是$10^{-100}$也是不等于零,但数值计算不一样,很多时候$10^{-10}$就可以当零看待。
因此,我们希望将秩的概念推广到更符合数值计算特性的形式,这便是有效秩(Effective Rank)概念的由来。
误差截断
需要指出的是,目前学术界对有效秩并没有统一的定义,接下来我们介绍的是一些从不同角度切入来定义有效秩的思路。对于实际问题,读者可以自行选择适合的定义来使用。
[...]
","description":"秩(Rank)是线性代数中的重要概念,它代表了矩阵的内在维度。然而,数学上对秩的严格定义,很多时候并不完全适用于数值计算场景,因为秩等于非零奇异值的个数,而数学上对“等于零”这件事的理解跟数值计算有所不同,数学上的“等于零”是绝对地、严格地等于零,哪怕是$10^{-100}$也是不等于零,但数值计算不一样,很多时候$10^{-10}$就可以当零看待。 因此,我们希望将秩的概念推广到更符合数值计算特性的形式,这便是有效秩(Effective Rank)概念的由来。\\n\\n误差截断\\n\\n需要指出的是,目前学术界对有效秩并没有统一的定义…","guid":"https://kexue.fm/archives/10847","author":"苏剑林","authorUrl":null,"authorAvatar":null,"insertedAt":"2025-04-10T14:51:54.164Z","publishedAt":"2025-04-10T13:34:00.693Z","media":null,"categories":null,"attachments":null,"extra":null,"language":null,"feeds":{"type":"feed","id":"52351163392512007","url":"https://kexue.fm/feed","title":"科学空间|Scientific Spaces","description":"渴望成为一个小飞侠","siteUrl":"https://kexue.fm/","image":null,"errorMessage":null,"errorAt":null,"ownerUserId":null}}]}')