Self-supervised learning (SSL) is an increasingly popular paradigm for representation learning. Recent methods can be classified as sample-contrastive, dimension-contrastive, or asymmetric network-based, with each family having its own approach to avoiding informational collapse. While dimension-contrastive methods converge to similar solutions as sample-contrastive methods, it can be empirically shown that some methods require more epochs of training to converge. Motivated by closing this divide, we present the objective function FroSSL which is both sample- and dimension-contrastive up to embedding normalization. FroSSL works by minimizing covariance Frobenius norms for avoiding collapse and minimizing mean-squared error for augmentation invariance. We show that FroSSL converges more quickly than a variety of other SSL methods and provide theoretical and empirical support that this faster convergence is due to how FroSSL affects the eigenvalues of the embedding covariance matrices. We also show that FroSSL learns competitive representations on linear probe evaluation when used to train a ResNet18 on the CIFAR-10, CIFAR-100, STL-10, and ImageNet datasets.

A machine learning model is traditionally considered robust if its prediction remains (almost) constant under input perturbations with small norm. However, real-world tasks like molecular property prediction or point cloud segmentation have inherent equivariances, such as rotation or permutation equivariance. In such tasks, even perturbations with large norm do not necessarily change an input's semantic content. Furthermore, there are perturbations for which a model's prediction explicitly needs to change. For the first time, we propose a sound notion of adversarial robustness that accounts for task equivariance. We then demonstrate that provable robustness can be achieved by (1) choosing a model that matches the task's equivariances (2) certifying traditional adversarial robustness. Certification methods are, however, unavailable for many models, such as those with continuous equivariances. We close this gap by developing the framework of equivariance-preserving randomized smoothing, which enables architecture-agnostic certification. We additionally derive the first architecture-specific graph edit distance certificates, i.e. sound robustness guarantees for isomorphism equivariant tasks like node classification. Overall, a sound notion of robustness is an important prerequisite for future work at the intersection of robust and geometric machine learning.

Spatial confounding poses a significant challenge in scientific studies involving spatial data, where unobserved spatial variables can influence both treatment and outcome, possibly leading to spurious associations. To address this problem, we introduce SpaCE: The Spatial Confounding Environment, the first toolkit to provide realistic benchmark datasets and tools for systematically evaluating causal inference methods designed to alleviate spatial confounding. Each dataset includes training data, true counterfactuals, a spatial graph with coordinates, and smoothness and confounding scores characterizing the effect of a missing spatial confounder. It also includes realistic semi-synthetic outcomes and counterfactuals, generated using state-of-the-art machine learning ensembles, following best practices for causal inference benchmarks. The datasets cover real treatment and covariates from diverse domains, including climate, health and social sciences. SpaCE facilitates an automated end-to-end pipeline, simplifying data loading, experimental setup, and evaluating machine learning and causal inference models. The SpaCE project provides several dozens of datasets of diverse sizes and spatial complexity. It is publicly available as a Python package, encouraging community feedback and contributions.

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. A Nigerian military attack that used drones to target rebels instead killed some civilians, government and military officials said Monday. The misfire during a religious celebration was the latest such errant bombing of local residents in Nigeria's violence hot spots. Muslims observing Maulud on Sunday night in Kaduna state's Igabi council area were "mistakenly killed and many others injured" by the drone "targeting terrorists and bandits," Gov. Uba Sani said.

Concerned Veterans for America and Air Force veteran Darin Selnick discusses Veteran Affairs facing scrutiny for medical care for migrants and the calls for Congress to investigate. House Republicans' campaign arm has rolled out a new political ad filled with images created using artificial intelligence (AI). The video runs just under a minute long and features AI-generated pictures depicting migrant encampments across some of the United States' most famous open-air landmarks. The political ad targets swing district Democrats near national parks. It comes after House Republicans passed a bill earlier this week that would ban the use of U.S. government funds going toward housing undocumented migrants on federally owned land.

FOX News White House correspondent Peter Doocy has the latest on the Biden administration's response to the Middle East conflict on'Special Report.' As Israeli Defense Forces resumed military operations to eradicate the Hamas terrorist threat last Friday, the Biden administration is inserting itself into Israel's war planning process, teaching the Israelis – who've been fighting for their survival for decades – how to properly prosecute the conflict. Washington warfare "experts" – who arguably haven't secured a single clear military victory since 1945 – insist that Israeli military strategists alter their war plans to make their combat operations more targeted and their strikes more accurate, in order to minimize casualties, especially among civilians. The Biden administration's demands, while noble-sounding, are misguided and unreasonable. Implementing these requirements, at the expense of achieving the main mission of eliminating Hamas and its entire supporting infrastructure, will likely prolong the conflict, ultimately resulting in many more Israeli and Palestinian deaths.

The United States and China have a relationship that could be summed up in a single word as, "complicated." And if you want to use more than one word, "kind of like that video of two dogs growling at each other through a gate." While megacorps just want to make as much money as possible, they have to keep this relationship in mind. Nvidia is in hot water with the US Commerce Department over recent chips designed specifically for the Chinese market. For context, Nvidia is making unbelievably, ridiculously, stupidly huge amounts of money at the moment, providing the hardware backbone for the AI software boom.

On December 2, the Israeli army's Arabic-language spokesperson Avichay Adraee posted a map of Gaza, broken up into a grid of numbered blocks with instructions that Palestinians living in certain areas evacuate to Rafah. Leaflets containing a QR code linking to the map on the Israeli army's website were also dropped over Gaza. This move came as Israeli fighter jets bombarded the south of the Strip – previously designated as a "safe zone" – killing hundreds of Palestinians in 24 hours. The Israeli army proudly announced that it had hit "400 targets". Meanwhile, media reports revealed that the Israeli army's ability to intensify what it calls "precision" air strikes has been boosted by an artificial intelligence (AI) tool that generates "targets".

One of the most stunning twists in the recent five-day crisis at ChatGPT creator OpenAI came when some 95 percent of the company's hundreds of employees threatened to quit. The staff planned to follow CEO Sam Altman to develop successors to ChatGPT at Microsoft instead. The threat appeared to mark a turning point in Altman's ultimately successful attempt to return to OpenAI--it was also a scenario that businesses have the legal power to block in most US states. California, home to OpenAI's San Francisco HQ, is one of a handful states that bar the enforcement of noncompete agreements in employment contracts, which can forbid employees from hopping jobs to a competitor, often for years. That picture is now set to change, as a raft of new legislation aims to make more places like California.

In October, Francesca Mani was one of reportedly more than 30 girls at Westfield High School in New Jersey who were victims of deepfake pornography. Boys at the school had taken photos of Francesca and her classmates and manipulated them with artificial intelligence to create sexually explicit images of them without their consent. The practice is actually stunningly commonplace, but we rarely hear such stories--at least in part because many victims of sexual harassment very understandably don't want to talk publicly about incidents that are so private. But within just a day of learning about the violation, which she calls "shocking," 15-year-old Francesca started speaking out and calling on lawmakers to do something about the broader problem. Her efforts are already starting to pay off with new momentum behind proposals for state and federal legislation, which I wrote about in a story published this morning.