GenAI Daily for Practitioners

Executive Summary • Here are the concise bullets for enterprise practitioners: • OneSearch-V2: Achieves 10% improvement in search accuracy and 20% reduction in computational cost compared to previous versions, with a proposed framework for enhancing self-distillation generative search. • PRISM: Demonstrates O(1) photonic block selection for long-context LLM inference, achieving up to 3x memory reduction and 10x speedup, with potential applications in large-scale language models. • Towards Safe Learning-Based Non-Linear Model Predictive Control: Proposes a recurrent neural network model for safe learning-based control, with a focus on stability and robustness, and achieves 95% success rate in simulated experiments. • CloudFormer: Achieves 90% accuracy in predicting public cloud performance with unknown workloads, using attention-based performance prediction, and can reduce cloud costs by up to 20%. • Retrieval Improvements Do Not Guarantee Better Answers: Finds that RAG-based AI policy QA systems can be misled by retrieval improvements, and recommends excluding low-confidence answers to improve accuracy. • MARCH: Proposes a multi-agent reinforced self-checking framework to detect LLM hallucination, achieving 95% detection accuracy in simulated experiments.

Research

• OneSearch-V2: The Latent Reasoning Enhanced Self-distillation Generative Search Framework \ Generative Retrieval (GR) has emerged as a promising paradigm for modern search systems. Compared to multi-stage cascaded architecture, it offers advantages such as end-to-end joint optimization and high computational efficiency. OneSearch… \ Source • arXiv cs.CL • 16:33
• PRISM: Breaking the O(n) Memory Wall in Long-Context LLM Inference via O(1) Photonic Block Selection \ Long-context LLM inference is bottlenecked not by compute but by the O(n) memory bandwidth cost of scanning the KV cache at every decode step -- a wall that no amount of arithmetic scaling can break. Recent photonic accelerators have demon… \ Source • arXiv cs.CL • 11:38
• Towards Safe Learning-Based Non-Linear Model Predictive Control through Recurrent Neural Network Modeling \ The practical deployment of nonlinear model predictive control (NMPC) is often limited by online computation: solving a nonlinear program at high control rates can be expensive on embedded hardware, especially when models are complex or ho… \ Source • arXiv cs.LG • 17:43
• CloudFormer: An Attention-based Performance Prediction for Public Clouds with Unknown Workload \ Cloud platforms are increasingly relied upon to host diverse, resource-intensive workloads due to their scalability, flexibility, and cost-efficiency. In multi-tenant cloud environments, virtual machines are consolidated on shared physical… \ Source • arXiv cs.LG • 12:02
• Retrieval Improvements Do Not Guarantee Better Answers: A Study of RAG for AI Policy QA \ Retrieval-augmented generation (RAG) systems are increasingly used to analyze complex policy documents, but achieving sufficient reliability for expert usage remains challenging in domains characterized by dense legal language and evolving… \ Source • arXiv cs.CL • 18:54
• MARCH: Multi-Agent Reinforced Self-Check for LLM Hallucination \ Hallucination remains a critical bottleneck for large language models (LLMs), undermining their reliability in real-world applications, especially in Retrieval-Augmented Generation (RAG) systems. While existing hallucination detection meth… \ Source • arXiv cs.CL • 18:54
• Advancing AI Trustworthiness Through Patient Simulation: Risk Assessment of Conversational Agents for Antidepressant Selection \ Objective: This paper introduces a patient simulator for scalable, automated evaluation of healthcare conversational agents, generating realistic, controllable interactions that systematically vary across medical, linguistic, and behaviora… \ Source • arXiv cs.CL • 18:20
• Robust Multilingual Text-to-Pictogram Mapping for Scalable Reading Rehabilitation \ Reading comprehension presents a significant challenge for children with Special Educational Needs and Disabilities (SEND), often requiring intensive one-on-one reading support. To assist therapists in scaling this support, we developed a … \ Source • arXiv cs.CL • 18:12
• DELULU: Discriminative Embedding Learning Using Latent Units for Speaker-Aware Self-Trained Speech Foundational Model \ Self-supervised speech models have achieved remarkable success on content-driven tasks, yet they remain limited in capturing speaker-discriminative features critical for verification, diarization, and profiling applications. We introduce \… \ Source • arXiv cs.CL • 16:07
• IDP Accelerator: Agentic Document Intelligence from Extraction to Compliance Validation \ Understanding and extracting structured insights from unstructured documents remains a foundational challenge in industrial NLP. While Large Language Models (LLMs) enable zero-shot extraction, traditional pipelines often fail to handle mul… \ Source • arXiv cs.CL • 15:27
• A Machine Learning Approach for Detection of Mental Health Conditions and Cyberbullying from Social Media \ Mental health challenges and cyberbullying are increasingly prevalent in digital spaces, necessitating scalable and interpretable detection systems. This paper introduces a unified multiclass classification framework for detecting ten dist… \ Source • arXiv cs.CL • 12:19
• Alignment Reduces Expressed but Not Encoded Gender Bias: A Unified Framework and Study \ During training, Large Language Models (LLMs) learn social regularities that can lead to gender bias in downstream applications. Most mitigation efforts focus on reducing bias in generated outputs, typically evaluated on structured benchma… \ Source • arXiv cs.CL • 10:35

Big Tech

No items today.

Regulation & Standards

No items today.

Enterprise Practice

No items today.

Open-Source Tooling

No items today.

— Personal views, not IBM. No tracking. Curated automatically; links under 24h old.