The indicator matrix plays an important role in machine learning, but optimizing it is an NP-hard problem. We propose a new relaxation of the indicator matrix and prove that this relaxation forms a manifold, which we call the Relaxed Indicator Matrix Manifold (RIM manifold). Based on Riemannian geometry, we develop a Riemannian toolbox for optimization on the RIM manifold. Specifically, we provide several methods of Retraction, including a fast Retraction method to obtain geodesics. We point out that the RIM manifold is a generalization of the double stochastic manifold, and it is much faster than existing methods on the double stochastic manifold, which has a complexity of \( \mathcal{O}(n^3) \), while RIM manifold optimization is \( \mathcal{O}(n) \) and often yields better results. We conducted extensive experiments, including image denoising, with millions of variables to support our conclusion, and applied the RIM manifold to Ratio Cut, we provide a rigorous convergence proof and achieve clustering results that outperform the state-of-the-art methods. Our Code in \href{https://github.com/Yuan-Jinghui/Riemannian-Optimization-on-Relaxed-Indicator-Matrix-Manifold}{here}.

This paper provides a practical introduction to Double/Debiased Machine Learning (DML). DML provides a general approach to performing inference about a target parameter in the presence of nuisance parameters. The aim of DML is to reduce the impact of nuisance parameter estimation on estimators of the parameter of interest. We describe DML and its two essential components: Neyman orthogonality and cross-fitting. We highlight that DML reduces functional form dependence and accommodates the use of complex data types, such as text data. We illustrate its application through three empirical examples that demonstrate DML's applicability in cross-sectional and panel settings.

The exploration of novel architectures requires physics-based simulation due to a lack of prior experience to start from, which introduces two specific challenges for optimization algorithms: evaluations become more expensive (in time) and evaluations might fail. The former challenge is addressed by Surrogate-Based Optimization (SBO) algorithms, in particular Bayesian Optimization (BO) using Gaussian Process (GP) models. An overview is provided of how BO can deal with challenges specific to architecture optimization, such as design variable hierarchy and multiple objectives: specific measures include ensemble infills and a hierarchical sampling algorithm. Evaluations might fail due to non-convergence of underlying solvers or infeasible geometry in certain areas of the design space. Such failed evaluations, also known as hidden constraints, pose a particular challenge to SBO/BO, as the surrogate model cannot be trained on empty results. This work investigates various strategies for satisfying hidden constraints in BO algorithms. Three high-level strategies are identified: rejection of failed points from the training set, replacing failed points based on viable (non-failed) points, and predicting the failure region. Through investigations on a set of test problems including a jet engine architecture optimization problem, it is shown that best performance is achieved with a mixed-discrete GP to predict the Probability of Viability (PoV), and by ensuring selected infill points satisfy some minimum PoV threshold. This strategy is demonstrated by solving a jet engine architecture problem that features at 50% failure rate and could not previously be solved by a BO algorithm. The developed BO algorithm and used test problems are available in the open-source Python library SBArchOpt.

Addressing multi-modality constitutes one of the major challenges of sampling. In this reflection paper, we advocate for a more systematic evaluation of samplers towards two sources of difficulty that are mode separation and dimension. For this, we propose a synthetic experimental setting that we illustrate on a selection of samplers, focusing on the challenging criterion of recovery of the mode relative importance. These evaluations are crucial to diagnose the potential of samplers to handle multi-modality and therefore to drive progress in the field.

There are reasonable arguments to suggest that AI tools may eventually help reduce emissions, as the IEA report underscores. But what we know for sure is that they're driving up energy demand and emissions today--especially in the regional pockets where data centers are clustering. So far, these facilities, which generally run around the clock, are substantially powered through natural-gas turbines, which produce significant levels of planet-warming emissions. Electricity demands are rising so fast that developers are proposing to build new gas plants and convert retired coal plants to supply the buzzy industry. The other thing we know is that there are better, cleaner ways of powering these facilities already, including geothermal plants, nuclear reactors, hydroelectric power, and wind or solar projects coupled with significant amounts of battery storage. The trade-off is that these facilities may cost more to build or operate, or take longer to get up and running.

This summer, The Sphere in Las Vegas is going to debut a new experience: The Wizard of Oz at Sphere. "The power of generative AI, combined with Google's infrastructure and expertise, is helping us to achieve something extraordinary," said Sphere Entertainment Executive Chairman and CEO Jim Dolan in a statement provided to Mashable. "We needed a partner who could push boundaries alongside our teams at Sphere Studios and Magnopus, and Google was the only company equipped to meet the challenge on the world's highest resolution LED screen." Regardless of whether you've been to Vegas, you're likely familiar with The Sphere. It's constantly going viral with its 580,000 square feet of LED displays wrapped around the venue.

The US Department of Justice has indicted Albert Sangier for defrauding investors with misleading statements about his Nate financial technology platform. Founded by Sangier in 2018, Nate claimed it could offer shoppers a universal checkout app thanks to artificial intelligence. However, the indictment states that the so-called AI-powered transactions in Nate were actually completed by human contractors in the Philippines and Romania or by bots. Sangier raised more than 40 million from investors for the app. This case follows reporting by The Information in 2022 that cast light on Nate's use of human labor rather than AI.

OpenAI has expanded ChatGPT's existing ability to remember more information about you. On Thursday, the AI company announced via X that "ChatGPT can now reference all of your past chats to provide more personalized responses." There was already a memory setting that, when toggled on, enabled ChatGPT to remember saved memories and reference them in conversations. But now ChatGPT can remember even more. "In addition to the saved memories that were there before, it can now reference your past chats to deliver responses that feel noticeably more relevant and useful," OpenAI explained in the X thread.

The video game industry's first direct-join union has grown to roughly 445 members since its launch, amidst industry-wide job losses and an escalating federal crackdown on workers' rights. The United Videogame Workers union, which launched with the Communications Workers of America (CWA), was announced March 19 at the Game Developers Conference. It's an effort on behalf of developers and the CWA to champion unionization efforts without relying on the National Labor Relations Board (NLRB), a federal agency that protects worker's rights and working conditions. Their first campaign will focus on industry-wide layoffs; a GDC report released in January found that 11 percent of developers surveyed said they'd been laid off in the year prior. The move comes at a time when the Trump administration has been hostile toward unions, issuing an executive order to end collective bargaining obligations with some federal agencies and firing an NLRB employee, crippling the agency.

Benchmark performance results typically accompany the launch of every new AI model to showcase how well the models can perform on various tasks. However, these tasks are not catered to individual industries but are more general, such as grade school mathematics (GSM8K) or graduate-level reasoning (GPQA). To fill that gap, OpenAI launched the OpenAI Pioneers Program, intended to advance AI model development for specific industries and real-world use cases. The program is a two-pronged effort in which companies will collaborate with OpenAI researchers to develop more domain-specific evaluations and fine-tuned models. In the blog post, OpenAI shared that "industries like legal, finance, insurance, healthcare, accounting, and many others are missing a unified source of truth for model benchmarking."