The first reports appeared on foreign screens, beyond the reach of most Iranians. On 28 February Prime Minister Benjamin Netanyahu said there were signs that the tyrant is no more, suggesting Supreme Leader Ayatollah Ali Khamenei had been killed in a joint US-Israeli strike. Iranians watching state television, however, encountered silence. Government officials would neither confirm nor deny Khamenei's death. On one of the state broadcaster's channels, IRTV3, one news presenter urged viewers to trust him and the latest information the government had.

While diffusion models have emerged as a powerful class of generative models, their learning dynamics remain poorly understood. We address this issue first by empirically showing that standard diffusion models trained on natural images exhibit a distributional simplicity bias, learning simple, pair-wise input statistics before specializing to higher-order correlations. We reproduce this behaviour in simple denoisers trained on a minimal data model, the mixed cumulant model, where we precisely control both pair-wise and higher-order correlations of the inputs. We identify a scalar invariant of the model that governs the sample complexity of learning pair-wise and higher-order correlations that we call the diffusion information exponent, in analogy to related invariants in different learning paradigms. Using this invariant, we prove that the denoiser learns simple, pair-wise statistics of the inputs at linear sample complexity, while more complex higher-order statistics, such as the fourth cumulant, require at least cubic sample complexity. We also prove that the sample complexity of learning the fourth cumulant is linear if pair-wise and higher-order statistics share a correlated latent structure. Our work describes a key mechanism for how diffusion models can learn distributions of increasing complexity.

Reinforcement learning (RL) has become a standard technique for post-training diffusion-based image synthesis models, as it enables learning from reward signals to explicitly improve desirable aspects such as image quality and prompt alignment. In this paper, we propose an online RL variant that reduces the variance in the model updates by sampling paired trajectories and pulling the flow velocity in the direction of the more favorable image. Unlike existing methods that treat each sampling step as a separate policy action, we consider the entire sampling process as a single action. We experiment with both high-quality vision language models and off-the-shelf quality metrics for rewards, and evaluate the outputs using a broad set of metrics. Our method converges faster and yields higher output quality and prompt alignment than previous approaches.

Random sample consensus (RANSAC), which is based on a repetitive sampling from a given dataset, is one of the most popular robust estimation methods. In this study, an energy-based model (EBM) for robust estimation that has a similar scheme to RANSAC, energy-based RANSAC (EB-RANSAC), is proposed. EB-RANSAC is applicable to a wide range of estimation problems similar to RANSAC. However, unlike RANSAC, EB-RANSAC does not require a troublesome sampling procedure and has only one hyperparameter. The effectiveness of EB-RANSAC is numerically demonstrated in two applications: a linear regression and maximum likelihood estimation.

Cyberbullying on social media is inherently multilingual and multi-faceted, where abusive behaviors often overlap across multiple categories. Existing methods are commonly limited by monolingual assumptions or single-task formulations, which restrict their effectiveness in realistic multilingual and multi-label scenarios. In this paper, we propose HMS-BERT, a hybrid multi-task self-training framework for multilingual and multi-label cyberbullying detection. Built upon a pretrained multilingual BERT backbone, HMS-BERT integrates contextual representations with handcrafted linguistic features and jointly optimizes a fine-grained multi-label abuse classification task and a three-class main classification task. To address labeled data scarcity in low-resource languages, an iterative self-training strategy with confidence-based pseudo-labeling is introduced to facilitate cross-lingual knowledge transfer. Experiments on four public datasets demonstrate that HMS-BERT achieves strong performance, attaining a macro F1-score of up to 0.9847 on the multi-label task and an accuracy of 0.6775 on the main classification task. Ablation studies further verify the effectiveness of the proposed components.

In this paper, we consider the problem of black-box optimization with noisy feedback revealed in batches, where the unknown function to optimize has a bounded norm in some Reproducing Kernel Hilbert Space (RKHS). We refer to this as the Batched Kernelized Bandits problem, and refine and extend existing results on regret bounds. For algorithmic upper bounds, (Li and Scarlett, 2022) shows that $B=O(\log\log T)$ batches suffice to attain near-optimal regret, where $T$ is the time horizon and $B$ is the number of batches. We further refine this by (i) finding the optimal number of batches including constant factors (to within $1+o(1)$), and (ii) removing a factor of $B$ in the regret bound. For algorithm-independent lower bounds, noticing that existing results only apply when the batch sizes are fixed in advance, we present novel lower bounds when the batch sizes are chosen adaptively, and show that adaptive batches have essentially same minimax regret scaling as fixed batches. Furthermore, we consider a robust setting where the goal is to choose points for which the function value remains high even after an adversarial perturbation. We present the robust-BPE algorithm, and show that a suitably-defined cumulative regret notion incurs the same bound as the non-robust setting, and derive a simple regret bound significantly below that of previous work.

See the full list of nominees and winners from the 98th Academy Awards, Hollywood's biggest night.

Pope Leo XIV called Sunday for an immediate ceasefire in the war involving Iran, marking the pontiff's strongest remarks yet on conflict.

Could Iran be using China's BeiDou system? Ukraine wants money and technology as payback after sending specialists to the Middle East to help down Iranian drones during the ongoing Israel-United States war with Iran . President Volodymyr Zelenskyy told reporters on Sunday that three teams were sent to the region to undertake expert assessments and demonstrate how drone defences work as countries in the Middle East continue to be targeted by Iran over hosting US military bases. We are not at war with Iran," Zelenskyy said. Earlier this week, Ukraine's leader announced military teams were sent to Qatar, the United Arab Emirates, Saudi Arabia, and a US military base in Jordan. But he explained that more long-term drone deals could be negotiated with Gulf countries, and what Kyiv gets in return for its assistance still needs to be established. "For us today, both the technology and the funding are important," Zelenskyy said. Throughout the four-year Russia-Ukraine war, Moscow has widely used Iranian Shahed-136 "suicide" drones, giving Kyiv expertise in knowing how to down the unmanned aerial vehicles through cheap drone interceptors, electronic jamming tools, and anti-aircraft weaponry. However, US President Donald Trump has said he does not need Ukraine's help in taking down Iranian drones attacking American targets. Zelenskyy said he doesn't know why Washington hasn't signed a drone agreement with Kyiv, which it has pushed for months. "I wanted to sign a deal worth about $35bn-50bn," he said. Still, as the Russia-Ukraine conflict continues with no end in sight, Zelenskyy raised concerns that the ongoing war in the Middle East will impact Kyiv's supplies of air defence missiles. "We would very much not like the United States to step away from the issue of Ukraine because of the Middle East," he told reporters. But as interest has grown for Ukrainian drone interceptors in light of the war, Zelenskyy said Kyiv's rules to buy the drones must be tightened, with foreign countries and firms being unable to bypass the government and talk directly to manufacturers. "Unfortunately, representatives of certain governments or companies want to bypass the Ukrainian state to purchase specific equipment," Zelensky told reporters. "Even in some free countries, we do not initially receive contracts from the private sector.

Two people have died following an outbreak of invasive meningitis at the University of Kent. BBC South East understands that a further 11 people from the Canterbury area are currently in hospital and reported to be seriously ill. It is understood that most are aged between 18 and 21 and are students at the university. Both of the people who have died are also believed to be between 18 and 21, with one also confirmed to be a student. More than 30,000 students, staff and their families are being contacted by the UK Health Security Agency (UKHSA) to inform them of the situation.