Author: Advanced AI Editor

The advent of large-scale vision foundation models, pre-trained on diverse natural images, has marked a paradigm shift in computer vision. However, how the frontier vision foundation models’ efficacies transfer to specialized domains remains such as medical imaging remains an open question. This report investigates whether DINOv3, a state-of-the-art self-supervised vision transformer (ViT) that features strong capability in dense prediction tasks, can directly serve as a powerful, unified encoder for medical vision tasks without domain-specific pre-training. To answer this, we benchmark DINOv3 across common medical vision tasks, including 2D/3D classification and segmentation on a wide range of medical imaging modalities. We…

Mistral AI fait son entrée dans la cour des grands. Mardi 9 septembre, la start-up française d’intelligence artificielle a annoncé avoir bouclé une levée de fonds d’un montant de 1,7 milliards d’euros. Grâce à ce nouveau tour de table, elle décroche le statut de décacorne, terme désignant une start-up dont la valorisation dépasse les 10 milliards d’euros. Mistral AI est en effet aujourd’hui valorisée à 11,7 milliards d’euros, contre 5,8 milliards d’euros au printemps 2024. Une première pour une entreprise de la French Tech.Cette levée de fonds marque également l’arrivée au capital de la start-up d’ASML, leader néerlandais des semi-conducteurs, dont l’investissement…

In early June, shortly after the beginning of the Atlantic hurricane season, Google unveiled a new model designed specifically to forecast the tracks and intensity of tropical cyclones. Part of the Google DeepMind suite of AI-based weather research models, the “Weather Lab” model for cyclones was a bit of an unknown for meteorologists at its launch. In a blog post at the time, Google said its new model, trained on a vast dataset that reconstructed past weather and a specialized database containing key information about hurricanes tracks, intensity, and size, had performed well during pre-launch testing. “Internal testing shows that…

Artificial intelligence is rapidly reshaping the landscape of customer service, offering businesses new ways to engage with consumers and streamline operations. From chatbots to predictive analytics, AI-driven solutions are now at the forefront of digital transformation, enabling organisations to deliver faster, more personalised support. As industries adapt, the implications for business models and consumer expectations are profound and far-reaching.The Expanding Role of AI in Customer EngagementThe integration of AI into customer service has moved well beyond simple automation. Today, intelligent systems can interpret natural language, understand context, and even anticipate customer needs before they arise. This shift allows companies to…

We are in an era of booming large models, with various industries eager to leverage their powerful capabilities. However, how to free these models from the constraints of ‘static’ learning and achieve lifelong learninghas become a necessary path toward achieving true AGI. Recently, researchers from the Massachusetts Institute of Technology (MIT) Improbable AILab published a research paper on arXivfocusing on the issue of catastrophic forgettingin post-training of large models, proposing a thought-provoking viewpoint: the Occam’s razor principlemay be key to solving this problem. SFT vs. RL: Who is Better at Knowledge Retention? The core of the research lies in comparing…

On the night of September 9, Tencent released and open-sourced the latest image model “Hunyuan Image 2.1”. This model boasts industry-leading capabilities and supports native 2K high-definition images.After being open-sourced, the Hunyuan Image 2.1 model quickly climbed the Hugging Face model popularity chart, becoming the third most popular model globally. Among the top eight models on the list, Tencent’s Hunyuan model family occupies three positions.At the same time, the Tencent Hunyuan team revealed that they will soon release a native multimodal image generation model.Hunyuan Image 2.1 is a comprehensive upgrade from the 2.0 architecture, placing greater emphasis on balancing generation…

Shutterstock Inc. SSTK, a well-known platform for images and videos, has entered into a partnership with Reka AI, a startup specializing in large language models. The financial terms were not disclosed.As part of this partnership, Reka will license Shutterstock’s extensive library of digital visual assets to assist in developing frontier-class multimodal language AI models.In addition to continued investment in human-created metadata, Shutterstock expects to leverage Reka’s multimodal models to augment details and improve the metadata attached to its library of digital assets.The licensing business with AI firms generated $104 million in revenue last year, CEO Paul J. Hennessy said in…

Deep research systems, agentic AI that solve complex, multi-step tasks by coordinating reasoning, search across the open web and user files, and tool use, are moving toward hierarchical deployments with a Planner, Coordinator, and Executors. In practice, training entire stacks end-to-end remains impractical, so most work trains a single planner connected to core tools such as search, browsing, and code. While SFT imparts protocol fidelity, it suffers from imitation and exposure biases and underuses environment feedback. Preference alignment methods such as DPO are schema and proxy-dependent, off-policy, and weak for long-horizon credit assignment and multi-objective trade-offs. A further limitation of…

AI disruption has sent shockwaves through the software sector this year. On Tuesday, a top OpenAI executive hammered home the potential threat to this lucrative industry.During a talk at a Goldman Sachs tech conference in San Francisco, OpenAI CFO Sarah Friar was asked about recent comments her boss made about the SaaS business.OpenAI CEO Sam Altman recently tweeted, “entering the fast fashion era of SaaS very soon.”Friar explained that the fast fashion phenomenon radically changed the apparel retail industry, and now the software-as-a-service business may be about to experience a similar disruption.”The whole value chain is getting disrupted. The people…

Mercor, a startup that connects companies like OpenAI and Meta with domain experts needed to train and refine their foundational AI models, is in discussions with investors for a Series C round, according to a marketing document viewed by TechCrunch and two sources familiar with the deal talks. Felicis, a returning investor, is considering doubling down on the company for the Series C, according to two sources. Felicis declined to comment.  The company is currently targeting a valuation of $10 billion or more, one person said. That’s up from an $8 billion target valuation that the company discussed a couple…