OpenAI has acquired Torch, a healthcare startup that aims to consolidate fragmented medical records into a single AI-accessible system. CNBC reports that all Torch employees are joining OpenAI, though the company has chosen not to disclose financial terms. At face value, the deal looks like a simple expansion of OpenAI’s health portfolio. However, that explanation sidesteps the real story. Centralizing health data inside a private AI ecosystem is not a step toward progress. It signals a future where […]

Ver mais

Yes, there’s open-source and free-to-use software for the fashion industry! When it comes to creating powerful digital tools for this purpose, there are many developers worldwide building programs available for download and use without needing a huge budget or a partnership with a famous brand name. Most of these projects include everything from 3D visioning, simulation of fabrics, managing production processes, and many more.  We have below a few examples of the innovations that have become accessible through […]

Ver mais

arXiv:2601.02521v1 Announce Type: new Abstract: Automated analysis of volumetric medical imaging on edge devices is severely constrained by the high memory and computational demands of 3D Convolutional Neural Networks (CNNs). This paper develops a lightweight computer vision framework that reconciles the efficiency of 2D detection with the necessity of 3D context by reformulating volumetric Computer Tomography (CT) data as sequential video streams. This video-viewpoint paradigm is applied to the time-sensitive task of Intracranial Hemorrhage (ICH) detection using the […]

Ver mais

Hi! I’m an exec at a University AI research club. We are trying to build a gpu cluster for our student body so they can have reliable access to compute, but we aren’t sure where to start. Our goal is to have a cluster that can be improved later on – i.e. expand it with more GPUs. We also want something that is cost effective and easy to set up. The cluster will be used for training ML […]

Ver mais

arXiv:2601.17016v1 Announce Type: new Abstract: With the incredible advancements in Large Language Models (LLMs), many people have started using them to satisfy their information needs. However, utilizing LLMs might be problematic for political issues where disagreement is common and model outputs may reflect training-data biases or deliberate alignment choices. To better characterize such behavior, we assess the stances of nine LLMs on 24 politically sensitive issues using five prompting techniques. We find that models often adopt opposing stances […]

Ver mais

Machine unlearning offers a practical alternative to avoid full model re-training by approximately removing the influence of specific user data. While existing methods certify unlearning via statistical indistinguishability from re-trained models, these guarantees do not naturally extend to model outputs when inputs are adversarially perturbed. In particular, slight perturbations of forget samples may still be correctly recognized by the unlearned model – even when a re-trained model fails to do so – revealing a novel privacy risk: information […]

Ver mais

arXiv:2601.16447v1 Announce Type: new Abstract: Large language models (LLMs) have demonstrated exceptional performance in reasoning tasks such as mathematics and coding, matching or surpassing human capabilities. However, these impressive reasoning abilities face significant challenges in specialized domains. Taking Go as an example, although AlphaGo has established the high performance ceiling of AI systems in Go, mainstream LLMs still struggle to reach even beginner-level proficiency, let alone perform natural language reasoning. This performance gap between general-purpose LLMs and domain […]

Ver mais

arXiv:2601.15395v1 Announce Type: new Abstract: User interactions with language models vary due to static properties of the user (trait) and the specific context of the interaction (state). However, existing persona datasets (like PersonaChat, PANDORA etc.) capture only trait, and ignore the impact of state. We introduce Chameleon, a dataset of 5,001 contextual psychological profiles from 1,667 Reddit users, each measured across multiple contexts. Using the Chameleon dataset, we present three key findings. First, inspired by Latent State-Trait theory, […]

Ver mais

arXiv:2207.05067v2 Announce Type: replace-cross Abstract: Pairwise causal background knowledge about the existence or absence of causal edges and paths is frequently encountered in observational studies. Such constraints allow the shared directed and undirected edges in the constrained subclass of Markov equivalent DAGs to be represented as a causal maximally partially directed acyclic graph (MPDAG). In this paper, we first provide a sound and complete graphical characterization of causal MPDAGs and introduce a minimal representation of a causal MPDAG. […]

Ver mais