Like a mirage speeding across the dusty desert outside Pueblo, Colorado, the first hydrogen-fuel-cell passenger train in the United States is getting warmed up on its test track. It will soon be shipped to Southern California, where it is slated to carry riders on San Bernardino County's Arrow commuter rail service before the end of the year. The best way to decarbonize railroads is the subject of growing debate among regulators, industry, and activists. The debate is partly technological, revolving around whether hydrogen fuel cells, batteries, or overhead electric wires offer the best performance for different railroad situations. In the insular world of railroading, this hydrogen-powered train is a Rorschach test.

Question answering over heterogeneous data requires reasoning over diverse sources of data, which is challenging due to the large scale of information and organic coupling of heterogeneous data. Various approaches have been proposed to address these challenges. One approach involves training specialized retrievers to select relevant information, thereby reducing the input length. Another approach is to transform diverse modalities of data into a single modality, simplifying the task difficulty and enabling more straightforward processing. In this paper, we propose HProPro, a novel program-based prompting framework for the hybrid question answering task. HProPro follows the code generation and execution paradigm. In addition, HProPro integrates various functions to tackle the hybrid reasoning scenario. Specifically, HProPro contains function declaration and function implementation to perform hybrid information-seeking over data from various sources and modalities, which enables reasoning over such data without training specialized retrievers or performing modal transformations. Experimental results on two typical hybrid question answering benchmarks HybridQA and MultiModalQA demonstrate the effectiveness of HProPro: it surpasses all baseline systems and achieves the best performances in the few-shot settings on both datasets.

Jason Allen, a video game designer in Pueblo, Colorado, spent roughly 80 hours working on his entry to the Colorado State Fair's digital arts competition. Judges awarded him first place, which came with a $300 prize. But when Allen posted about his win on social media late last month, his artwork went viral--for all the wrong reasons. Allen's victory took a turn when he revealed online that he'd created his prize-winning art using Midjourney, an artificial intelligence program that can turn text descriptions into images. He says he also made that clear to state fair officials when he dropped off his submission, called Théâtre D'opéra Spatial.

The winning artwork was created using the AI tool Midjourney – which turns lines of text into astonishingly realistic graphics. The award came with a $300 cash prize. AI tools to generate images have been around for years with companies such as Google and OpenAI being notable investors in these text-to-image systems. "I'm not going to apologise for it … I won and I didn't break any rules," Allen, who is from Pueblo, Colorado, told The New York Times newspaper in an interview published on Friday. However, many have taken to social media to express their anger and despair over the award, arguing it took away from the hard work invested by humans to physically create noteworthy art.

Distributionally robust optimization (DRO) problems are increasingly seen as a viable method to train machine learning models for improved model generalization. These min-max formulations, however, are more difficult to solve. We therefore provide a new stochastic gradient descent algorithm to efficiently solve this DRO formulation. Our approach applies gradient descent to the outer minimization formulation and estimates the gradient of the inner maximization based on a sample average approximation. The latter uses a subset of the data in each iteration, progressively increasing the subset size to ensure convergence. Theoretical results include establishing the optimal manner for growing the support size to balance a fundamental tradeoff between stochastic error and computational effort. Empirical results demonstrate the significant benefits of our approach over previous work, and also illustrate how learning with DRO can improve generalization.