Articles for category: AI Research

March 14, 2025

ikayaniaamirshahzad@gmail.com

Curve Circuits

Author Contributions As we mentioned in Curve Detectors, our first investigation into curve neurons, it’s hard to separate author contributions between different papers in the Circuits project. Much of the original research on curve neurons came before we decided to separate the publications into the behavior of curve neurons and how they are built. In this section we’ve tried to isolate contributions specific to the mechanics of the curve neurons. Interface Design & Prototyping. Many weight diagrams were first prototyped by Chris during his first investigations of different families of neurons in early early vision, and some of these were

March 14, 2025

ikayaniaamirshahzad@gmail.com

Orthogonal Bases for Equivariant Graph Learning with Provable k-WL Expressive Power

Orthogonal Bases for Equivariant Graph Learning with Provable k-WL Expressive Power Jia He, Maggie Cheng; 26(29):1−35, 2025. Abstract Graph neural network (GNN) models have been widely used for learning graph-structured data. Due to the permutation-invariant requirement of graph learning tasks, a basic element in graph neural networks is the invariant and equivariant linear layers. Previous work (Maron et al., 2019b) provided a maximal collection of invariant and equivariant linear layers and a simple deep neural network model, called k-IGN, for graph data defined on k-tuples of nodes. It is shown that the expressive power of k-IGN is at least as

March 14, 2025

ikayaniaamirshahzad@gmail.com

Free Form Least-Squares Concept Erasure Without Oracle Concept Labels

This post assumes some familiarity with the idea of concept erasure and our LEACE concept erasure method. We encourage the reader to consult our arXiv paper for background. In our paper LEACE: Perfect linear concept erasure in closed form, we derived a concept erasure method that is least squares optimal within the class of affine transformations. We now extend this result by deriving the least squares optimal edit only under the restriction that the edited representation is a function of the unedited representation. In a previous blog post, we solved this problem in the case where the transformation may depend

March 14, 2025

ikayaniaamirshahzad@gmail.com

Stanford CRFM

Today, we are calling for AI developers to invest in the needs of third-party, independent researchers, who investigate flaws in AI systems. Our new paper advocates for a new standard of researcher protections, reporting and coordination infrastructure. The paper, In House Evaluation Is Not Enough: Towards Robust Third-Party Flaw Disclosure for General-Purpose AI, has 34 authors with expertise in machine learning, law, security, social science, and policy. Introduction Today, we are calling for AI developers to invest in the needs of third-party, independent researchers, who investigate flaws in AI systems. Our new paper advocates for a new standard of researcher

March 14, 2025

ikayaniaamirshahzad@gmail.com

Asymmetric Certified Robustness via Feature-Convex Neural Networks – The Berkeley Artificial Intelligence Research Blog

Asymmetric Certified Robustness via Feature-Convex Neural Networks TLDR: We propose the asymmetric certified robustness problem, which requires certified robustness for only one class and reflects real-world adversarial scenarios. This focused setting allows us to introduce feature-convex classifiers, which produce closed-form and deterministic certified radii on the order of milliseconds. Figure 1. Illustration of feature-convex classifiers and their certification for sensitive-class inputs. This architecture composes a Lipschitz-continuous feature map $\varphi$ with a learned convex function $g$. Since $g$ is convex, it is globally underapproximated by its tangent plane at $\varphi(x)$, yielding certified norm balls in the feature space. Lipschitzness of $\varphi$

March 14, 2025

ikayaniaamirshahzad@gmail.com

Experiment with Gemini 2.0 Flash native image generation

In December we first introduced native image output in Gemini 2.0 Flash to trusted testers. Today, we’re making it available for developer experimentation across all regions currently supported by Google AI Studio. You can test this new capability using an experimental version of Gemini 2.0 Flash (gemini-2.0-flash-exp) in Google AI Studio and via the Gemini API. Gemini 2.0 Flash combines multimodal input, enhanced reasoning, and natural language understanding to create images. Here are some examples of where 2.0 Flash’s multimodal outputs shine: 1. Text and images together Use Gemini 2.0 Flash to tell a story and it will illustrate it

March 14, 2025

ikayaniaamirshahzad@gmail.com

[2311.01797] On the Generalization Properties of Diffusion Models

[Submitted on 3 Nov 2023 (v1), last revised 12 Mar 2025 (this version, v4)] View a PDF of the paper titled On the Generalization Properties of Diffusion Models, by Puheng Li and 3 other authors View PDF HTML (experimental) Abstract:Diffusion models are a class of generative models that serve to establish a stochastic transport map between an empirically observed, yet unknown, target distribution and a known prior. Despite their remarkable success in real-world applications, a theoretical understanding of their generalization capabilities remains underdeveloped. This work embarks on a comprehensive theoretical exploration of the generalization attributes of diffusion models. We establish

March 14, 2025

ikayaniaamirshahzad@gmail.com

[2411.07223] Grounding Video Models to Actions through Goal Conditioned Exploration

[Submitted on 11 Nov 2024 (v1), last revised 12 Mar 2025 (this version, v2)] View a PDF of the paper titled Grounding Video Models to Actions through Goal Conditioned Exploration, by Yunhao Luo and 1 other authors View PDF HTML (experimental) Abstract:Large video models, pretrained on massive amounts of Internet video, provide a rich source of physical knowledge about the dynamics and motions of objects and tasks. However, video models are not grounded in the embodiment of an agent, and do not describe how to actuate the world to reach the visual states depicted in a video. To tackle this

March 13, 2025

ikayaniaamirshahzad@gmail.com

A Spanish Multi-Genre Dataset with Causal Relationships

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Source link