Subset Selection (MISS) problem, which aims to identify a subset of training samples with the greatest collective influence.

Safety and trustworthiness are indispensable requirements for real-world applications of AI systems using large language models (LLMs).

Is the Rat War Over? In New York, a rat czar and new methods have brought down complaints. We may even be ready to appreciate the creatures. Rats were leaving Manhattan, hurrying across the bridges in single-file lines. Some went to Westchester, some to Brooklyn. It was the pandemic, and the rats, which had been living off the nourishing trash of New York's densest borough for generations, were as panicked about the closure of restaurants as we were. People were eating three meals a day at home, and the rats were hungry. At least that was the story going around.

Thousands of Companies Are Driving China's AI Boom. How the Cyberspace Administration of China inadvertently made a guide to the country's homegrown AI revolution. When DeepSeek burst onto the global stage in January 2025, it seemed to appear out of nowhere. But the large language model was just one of the thousands of generative AI tools that have been released in China since 2023--and there's a public archive of every single one of them. Here are 23 ways China is rewiring the future .

If you want to put people back on the moon, don't gut the agency in charge of getting them there. The senator wanted a promise. For the last six years--or maybe the last decade or quarter century, depending on how you count it--the United States and China had been locked in a space race, a contest to see which nation could put its people on the moon . Senator Ted Cruz wanted President Donald Trump's nominee to run NASA, Jared Isaacman, to pledge that the US would not lose. Cruz brought a little surprise to Isaacman's confirmation hearing last April. It was a poster of the moon. On one side stood three astronauts and a giant Chinese flag. On the other were two more figures in space suits, with the tiniest Stars and Stripes planted in the lunar soil . Cruz apologized for the imbalance. "My team used ChatGPT," explained the senator, who chairs the committee that oversees NASA. Then Cruz, with a bit more seriousness, asked Isaacman, "Do we have your commitment that you will not allow the scenario on the right of this poster to happen? That China will not beat us to the moon?" Isaacman, a billionaire entrepreneur who had paid for his own missions to space, replied, "Senator, I only see the left-hand portion of that poster."

The recently proposed Conformer model has become the de facto backbone model for various downstream speech tasks based on its hybrid attention-convolution architecture that captures both local and global features. However, through a series of systematic studies, we find that the Conformer architecture's design choices are not optimal. After re-examining the design choices for both the macro and micro-architecture of Conformer, we propose Squeezeformer which consistently outperforms the state-of-the-art ASR models under the same training schemes. In particular, for the macro-architecture, Squeezeformer incorporates (i) the Temporal U-Net structure which reduces the cost of the multi-head attention modules on long sequences, and (ii) a simpler block structure of multi-head attention or convolution modules followed up by feed-forward module instead of the Macaron structure proposed in Conformer. Furthermore, for the micro-architecture, Squeezeformer (i) simplifies the activations in the convolutional block, (ii) removes redundant Layer Normalization operations, and (iii) incorporates an efficient depthwise down-sampling layer to efficiently sub-sample the input signal. Squeezeformer achieves state-of-the-art results of 7.5%, 6.5%, and 6.0% word-error-rate (WER) on LibriSpeech test-other without external language models, which are 3.1%, 1.4%, and 0.6% better than Conformer-CTC with the same number of FLOPs. Our code is open-sourced and available online.

Breakthroughs, discoveries, and DIY tips sent every weekday. It's been found in ancient human feces . The U.S. government stored 6.4 metric tons of it in mountains . And a big hunk of it played a major role in a presidential farewell party . While too much of the popular dairy product can spell tummy troubles and high cholesterol for some, new research suggests that eating more high-fat cheese and cream could be linked to a lower risk of developing dementia .

We present a rigorous information-geometric framework, grounded in diffusion dynamics, to quantify hallucinations in MLLMs where model outputs are embedded via spectral decompositions of multimodal graph Laplacians, and their gaps to a truth manifold define a semantic distortion metric. We derive Courant-Fischer bounds on a temperature-dependent hallucination profile and use RKHS eigen-modes to obtain modality-aware, interpretable measures that track evolution over prompts and time.

The rise of space AI is reshaping government and industry through applications such as disaster detection, border surveillance, and climate monitoring, powered by massive data from commercial and governmental low Earth orbit (LEO) satellites. Federated satellite learning (FSL) enables joint model training without sharing raw data, but suffers from slow convergence due to intermittent connectivity and introduces critical trust challenges--where biased or falsified updates can arise across satellite constellations, including those injected through cyberattacks on inter-satellite or satellite-ground communication links. We propose OrbitChain, a blockchain-backed framework that empowers trustworthy multi-vendor collaboration in LEO networks. OrbitChain (i) offloads consensus to high-altitude platforms (HAPs) with greater computational capacity, (ii) ensures transparent, auditable provenance of model updates from different orbits owned by different vendors, and (iii) prevents manipulated or incomplete contributions from affecting global FSL model aggregation. Extensive simulations show that OrbitChain reduces computational and communication overhead while improving privacy, security, and global model accuracy. Its permissioned proof-of-authority ledger finalizes over 1000 blocks with sub-second latency (0.16,s, 0.26,s, 0.35,s for 1-of-5, 3-of-5, and 5-of-5 quorums). Moreover, OrbitChain reduces convergence time by up to 30 hours on real satellite datasets compared to single-vendor, demonstrating its effectiveness for real-time, multi-vendor learning. Our code is available at https://github.com/wsu-cyber-security-lab-ai/OrbitChain.git