News

The US National Institute of Standards and Technology (NIST) has recently made a huge breakthrough in the security of predictive and generative AI systems. In a collaborative paper titled "Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations," Apostol Vassilev, a computer scientist at NIST, and his colleagues from Northeastern University and Robust…

Magic: The Gathering fans are abuzz with excitement over the latest promotional image conjured up by Wizards of the Coast! The artwork showcases a new range of cards in a retro frame, but some eagle-eyed observers have zoomed in on parts of the artwork to reveal what could be evidence of AI-generated content. This goes…

We are absolutely thrilled to announce the release of a comprehensive study into knowledge editing for Large Language Models (LLMs)! This paper, a collaboration between researchers from Zhejiang University, the National University of Singapore, the University of California, Ant Group, and Alibaba Group, is a remarkable breakthrough in Artificial Intelligence (AI) research. GPT-4 and other…

Diffusion models are advancing rapidly in the field of generative models, particularly for image generation, and are proving to be a cornerstone in the progress of artificial intelligence and machine learning. These models convert pure noise into detailed images through a denoising process, and have become increasingly important in computer vision and related fields. However,…

Today is a monumental day in the history of AI technology! On January 6th, 2024, Microsoft, OpenAI, and Google face legal challenges due to their artificial intelligence training methods. Authors and journalists, including Nicholas Basbanes and Nicholas Gage, have filed a lawsuit in Manhattan federal court against OpenAI, claiming that their books were used to…

Data-driven decision making is essential in our modern world, and the challenge of combining various data types such as images, tables, and text to extract meaningful insights can be daunting. Many researchers and professionals experience this issue when trying to forecast health outcomes using MRI scans and clinical data. Fortunately, Fusilli has emerged as a…

Be excited for the next advancement in the world of AI! Recent research has highlighted the success of Large Language Models (LLMs) trained on Code, showing excellence in diverse software engineering tasks. These models can be categorized into three main paradigms: (i) Code LLMs specialized in code completion, (ii) Task-specific Code LLMs fine-tuned for individual…

We are witnessing a remarkable breakthrough in language-related machine learning tasks, with the most impressive example being ChatGPT, which excels in complex language processing tasks. However, many mainsteam language learning models like LLaMA are pre-trained on English-dominant corpus and LaMDA, proposed by Google, is pre-trained on text containing over 90% English, limiting its performance in…

We are in for a revolutionary change in the Artificial Intelligence (AI) domain in the coming years! A team of researchers from the University of Illinois Urbana-Champaign have recently published a research paper that explores the powerful relationship between code and Large Language Models (LLMs). This remarkable study has opened up a world of possibilities…

Deep learning is revolutionizing various fields; plants are no exception! The recently introduced Eff-3DPSeg framework is a breakthrough in 3D plant shoot segmentation, leveraging annotation-efficient deep learning to overcome the challenges of expensive and time-consuming labeling processes. Using a Multi-view Stereo Pheno Platform (MVSP2) and Meshlab-based Plant Annotator (MPA), the researchers constructed a high-resolution point…

Behold the revolutionary self-play fine-tuning method, SPIN! Pioneered by researchers from UCLA, SPIN has ushered in a new era in the field of Artificial Intelligence (AI) through its natural language processing capabilities. This remarkable approach has the potential to convert a weak Large Language Model (LLM) to a strong one, without the need for any…

The legal battle surrounding the use of AI models in training off of copyrighted material has been heating up, and it's been revealed that over 16,000 artist names have been linked with the non-consensual training of Midjourney's image generation models! This disclosure was partially made in a 2023 class-action lawsuit and also in a recently…