The integration of external knowledge through Retrieval-Augmented Generation (RAG) has become foundational in enhancing large language models (LLMs) for knowledge-intensive tasks. However, existing RAG paradigms often overlook the cognitive step of applying knowledge, leaving a gap between retrieved facts and task-specific reasoning. In this work, we introduce RAG+, a principled and modular extension that explicitly incorporates application-aware reasoning into the RAG pipeline. RAG+ constructs a dual corpus consisting of knowledge and aligned application examples, created either manually or automatically, and retrieves both jointly during inference. This design enables LLMs not only to access relevant information but also to apply it within structured, goal-oriented reasoning processes. Experiments across mathematical, legal, and medical domains, conducted on multiple models, demonstrate that RAG+ consistently outperforms standard RAG variants, achieving average improvements of 3-5%, and peak gains up to 13.5% in complex scenarios. By bridging retrieval with actionable application, RAG+ advances a more cognitively grounded framework for knowledge integration, representing a step toward more interpretable and capable LLMs.

I have read the paper fast classification rates for high-dimensional conditional Gaussian models". The paper studies the problem of binary classification using a Gaussian model and provides some theoretical results on the convergence of the classification error rates (compared to the Bayes classifier). The paper presents some nice theoretical results and is interesting to some extent. I am generally positive about the paper but I have the following concerns. First, it is about the practical relevance.

Before this, fragmented computer networks struggled to communicate seamlessly. The introduction of the Transmission Control Protocol/Internet Protocol (TCP/IP) enabled consistent data transfer and became the standard for digital communication. However, this node-centric approach, which relies heavily on Internet Protocol (IP) addresses, has also created significant security vulnerabilities and privacy concerns due to its focus on network nodes rather than the data itself. In today's digital landscape, the centralized aggregation and storage of sensitive user data -- including IP addresses -- by service providers pose substantial security risks. These centralized repositories are prime targets for cyberattacks, potentially compromising user privacy and exposing sensitive information. Additionally, the reliance on IP-based system modeling has amplified these risks, necessitating a shift toward a more secure and resilient design approach. This paper proposes a novel data-centric design methodology that moves away from traditional node-focused models. By prioritizing data as the central entity and incorporating multimodal frameworks encompassing objects, events, concepts, and actions, this approach enhances data security and flexibility. The new informatics domain model reimagines data's role in system design, emphasizing its importance throughout its entire lifecycle to foster innovation, security, and seamless data interoperability.

Machine learning enables computers to learn from large amounts of data without being explicitly programmed to do so. We can already see how machine learning gives rise to new intelligent applications, from self-driving cars to intelligent assistants on our smartphones. Increasingly, businesses recognize the importance of using machine learning to transform their data assets into business value. However, many companies are unsure how machine learning can be applied to solve problems in an enterprise context. As the world's most relevant enterprise data is part of SAP's system and business network, SAP aspires to make all its enterprise solutions intelligent and help customers to leverage their data.

Machine learning enables computers to learn from large amounts of data without being explicitly programmed to do so. We can already see how machine learning gives rise to new intelligent applications, from self-driving cars to intelligent assistants on our smartphones. Increasingly, businesses recognize the importance of using machine learning to transform their data assets into business value. However, many companies are unsure how machine learning can be applied to solve problems in an enterprise context. As the world's most relevant enterprise data is part of SAP's system and business network, SAP aspires to make all its enterprise solutions intelligent and help customers to leverage their data.

Machine learning enables computers to learn from large amounts of data without being explicitly programmed to do so. We can already see how machine learning gives rise to new intelligent applications, from self-driving cars to intelligent assistants on our smartphones. Increasingly, businesses recognize the importance of using machine learning to transform their data assets into business value. However, many companies are unsure how machine learning can be applied to solve problems in an enterprise context. As the world's most relevant enterprise data is part of SAP's system and business network, SAP aspires to make all its enterprise solutions intelligent and help customers to leverage their data.