#machine-learning

#machine-learning

The study explores using large language models for efficient text classification through pseudo-labeled datasets.

Reliability in text classification is crucial and can be assessed using multiple annotators and LLMs to align with human consensus.

TnT-LLM framework enhances text classification using LLM for taxonomy generation and training lightweight classifiers with pseudo-labels from the generated taxonomy.

The article introduces a new evaluation suite for taxonomy generation and text classification using a combination of evaluation strategies.

OpenAI's new models, o3 and o4-mini, incorporate visual reasoning, enhancing AI understanding of images alongside text.

OpenAI's GPT-4.1 model significantly improves coding capabilities and supports long input prompts for better contextual processing.

OpenAI launched new AI models optimized for coding to compete with Google and Anthropic.

Generic machine learning applications are ruled ineligible for patents unless they show model improvements.

The Federal Circuit upheld the invalidation of Recentive's patent claims, deeming them unpatentable due to their generic application of machine learning.

Microsoft has developed the first 1-bit large language model, BitNet b1.58 2B4T, demonstrating significant computational efficiency.

Microsoft's BitNet b1.58 2B4T is the largest-scale 1-bit AI model capable of running on CPUs, offering impressive speed and efficiency.

AI simulates human intelligence in machines, allowing for advanced problem-solving and decision-making.

The study presents a new NLP pipeline that efficiently constructs knowledge graphs from unstructured scientific texts by fine-tuning language models with minimal data.

Organizations are shifting from manual data labeling to automated methods using Large Language Models (LLMs) to enhance efficiency and accuracy.

Effective communication between business leaders and ML practitioners is key to the success of machine learning projects.

The article discusses the training and evaluation of various LLM models with a focus on NER and internal relation recognition.

Apple has found a method to improve AI without accessing user data.

New training utilizes synthetic data compared to real user samples for better AI outputs.

Anthropic's research aims to enhance the interpretability of large language models by using a novel AI microscope approach.

Wikimedia is releasing a dataset for AI training to curb Wikipedia scraping.

Partnership with Kaggle aims to provide structured data for AI model training.

The dataset includes well-structured information while reducing server load from AI bots.

AI system Dreamer can play Minecraft and perform complex tasks without prior training, achieving human-level skill in diamond mining.

The iterative contrastive self-improvement method significantly enhances policy training efficiency and output quality.

The article discusses an experiment on Direct Nash Optimization methodologies using reinforcement learning from human feedback (RLHF) for preference modeling.

AI is transforming software development by shifting the focus from code to data.

Apache Spark is crucial for efficiently processing large datasets in modern enterprises.

The ODSC East 2025 event will showcase AI innovations from leading companies, focusing on practical applications and cutting-edge technologies.

Onehouse's Open Engines capability breaks the link between compute and storage, enhancing open data utilization.

Advanced agents like Simular AI's S2 demonstrate significant progress in automating computer-related tasks, surpassing previous models in performance.

Marketers should critically assess AI tools in martech to distinguish genuine innovation from rebranded automation.

Apple's strategy aims to enhance AI by analyzing user data on-device while prioritizing privacy.

OpenAI launched GPT-4.1, improving coding, instruction following, and long context comprehension.

New models support up to 1 million tokens and offer lower operational costs.

AI's influence in science is rapidly growing, but it necessitates cautious application to avoid overestimations and misuse.

AI's role in art raises questions about originality in creativity and expression.

David Salle's paintings blend traditional methods with AI-generated elements, prompting discourse on artistic authenticity.

Large language models excel due to their size and number of parameters, but smaller models can be equally effective for specific tasks.

Large language models (LLMs) and small language models (SLMs) will have distinct roles in generative AI applications.

Achieving precision in data retrieval is crucial for the operationalization of generative AI.

ODSC East 2025 offers students and professionals immersive, hands-on training in various AI disciplines designed for real-world application.

Robotic hands can learn manipulation tasks effectively, even without tactile sensation, when trained with a proper curriculum.

Picky robot engages users in recycling education using humor and interactive features.