In this work we study the robustness to adversarial attacks, of early-stopping strategies on gradient-descent (GD) methods for linear regression. More precisely, we show that early-stopped GD is optimally robust (up to an absolute constant) against Euclidean-norm adversarial attacks. However, we show that this strategy can be arbitrarily sub-optimal in the case of general Mahalanobis attacks. This observation is compatible with recent findings in the case of classification~\cite{Vardi2022GradientMP} that show that GD provably converges to non-robust models. To alleviate this issue, we propose to apply instead a GD scheme on a transformation of the data adapted to the attack. This data transformation amounts to apply feature-depending learning rates and we show that this modified GD is able to handle any Mahalanobis attack, as well as more general attacks under some conditions. Unfortunately, choosing such adapted transformations can be hard for general attacks. To the rescue, we design a simple and tractable estimator whose adversarial risk is optimal up to within a multiplicative constant of 1.1124 in the population regime, and works for any norm.

While Artificial Intelligence (AI) technologies are progressing fast, compliance costs have become a huge financial burden for AI startups, which are already constrained on research & development budgets. This situation creates a compliance trap, as many AI startups are not financially prepared to cope with a broad spectrum of regulatory requirements. Particularly, the complex and varying regulatory processes across the globe subtly give advantages to well-established and resourceful technology firms over resource-constrained AI startups [1]. The continuation of this trend may phase out the majority of AI startups and lead to giant technology firms' monopolies of AI technologies. To demonstrate the reality of the compliance trap, from a field deployment perspective, we delve into the details of compliance costs of AI commercial operations.

Not all contracts are good, but all good contracts can be expressed as a finite-state transition system ("State-Transition Contracts"). Contracts that can be represented as State-Transition Contracts discretize fat-tailed risk to foreseeable, managed risk, define the boundary of relevant events governed by the relationship, and eliminate the potential of inconsistent contractual provisions. Additionally, State-Transition Contracts reap the substantial benefit of being able to be analyzed under the rules governing the science of the theory of computation. Simple State-Transition Contracts can be represented as discrete finite automata; more complicated State-Transition Contracts, such as those that have downstream effects on other agreements or complicated pathways of performance, benefit from representation as weighted finite-state transducers, with weights assigned as costs, penalties, or probabilities of transitions. This research paper (the "Research" or "Paper") presents a complex legal transaction represented as weighted finite-state transducers. Furthermore, we show that the mathematics/algorithms permitted by the algebraic structure of weighted finite-state transducers provides actionable, legal insight into the transaction.

Imagine that you are teaching a technical subject to children in a small village. They are eager to learn, but you face a problem: There are few resources to educate them in their mother tongue. This is a common experience in India, where the quality of textbooks written in many local languages pales in comparison to those written in English. To address educational inequality, the Indian government launched an initiative in 2020 that would improve the quality of these resources for hundreds of millions of people, but its implementation remains a massive undertaking. Siddhartha Jayanti, an MIT PhD student in electrical engineering and computer science (EECS) who is an affiliate of MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) and Google Research, encountered this problem first-hand when teaching students in India about math, science, and English.

What can marketers bring to the mix when AI is so powerful? Don't miss a MarketingProfs podcast, subscribe to our free newsletter! Passion, for one thing, says Peter Prodromou of Boathouse. "If you're in the upper right-hand corner with passion, chances are people are going to want to work with you or buy your product," he says on the latest episode of Marketing Smarts. "Think about Apple and Tesla; those are two brands that are very much about passion. Your ability to convey that is critically important." AI is just an algorithm, after all. "Everybody is going to shop at Amazon because they have the best algorithm, and there may or may not be passion for it," Peter says.

Ukraine's Prosecutor General Andriy Kostin details Ukrainian efforts to investigate war crimes and responds to concerns of corruption within his country. Ukraine's military is planning to invest nearly $550 million in drones as the country mulls its next moves in its ongoing fight against Russia, according to a Monday report. Defense Minister Oleksii Reznikov said Ukraine is increasing the procurement of Unmanned Aerial Vehicles (UAVs), or drones, for reconnaissance and assault purposes this year and plans to allocate around $547.05 million for this segment. FILE: Ukrainian Defence Minister Oleksiy Reznikov attends a meeting of the Ukraine Defence contact group as part of a NATO Defence Ministers Council at the Alliance headquarters in Brussels on October 12, 2022. In the 11 months of fighting, Ukraine has received significant supplies of UAVS from its partners, including missile-equipped Bayraktar TB2's from Turkey and a Black Hornet reconnaissance drone from Norway.

Russia has begun its "big revenge" for Ukraine's resistance to its invasion, Ukrainian President Volodymyr Zelenskyy has said, as Russian forces claimed a series of incremental gains in his country's east. Zelenskyy has been warning for weeks that Moscow aims to step up its assault on Ukraine after about two months of virtual stalemate along the front line that stretches across the south and east. While there was no sign of a broader new offensive on Monday, the administrator of Russian-controlled parts of Ukraine's eastern Donetsk province, Denis Pushilin, said Russian troops had secured a foothold in Vuhledar, a coal mining town whose ruins have been a Ukrainian bastion since the outset of the war. Pushilin's adviser, Yan Gagin, said fighters from Russian mercenary force Wagner had taken partial control of a supply road leading to Bakhmut, a city that has been the focus of a Russian offensive for months. A day earlier, the head of Wagner said his fighters had secured Blahodatne, a village just north of Bakhmut.

The world faces a significant risk of passing a crucial global warming threshold earlier than scientists had suggested, possibly as soon as 2050, a paper published Monday found. The threshold is the point at which Earth's overall temperature has increased by 2.0 degrees Celsius, or 3.6 degrees Fahrenheit. If greenhouse gas emissions remain at high levels, there's a 50% probability the world could reach that catastrophic milestone before 2050, said co-author Noah Diffenbaugh, a climate scientist at Stanford University. The chance it could reach 2.0 degrees before 2058 is 84 in 100. Earlier estimates had put it closer to the end of the century.

Cybersecurity is a critical aspect of modern society, and it's only becoming more important as our lives become increasingly dependent on technology. As cyber threats evolve, it's crucial for organizations to have access to the latest tools and techniques for defending against them. That's where ChatGPT comes in. ChatGPT is a large language model developed by OpenAI that has been trained on a vast amount of data, making it a powerful tool for solving complex cybersecurity problems. By using advanced machine learning algorithms, ChatGPT can analyze large amounts of data in real-time and identify potential threats before they can cause harm. One example of how ChatGPT can be used in cybersecurity is to help detect and respond to phishing attacks.

A drone attack on an Iranian military facility that resulted in a large explosion in the center of the city of Isfahan on Saturday was the work of the Mossad, Israel's premier intelligence agency, according to senior intelligence officials who were familiar with the dialogue between Israel and the United States about the incident. The facility's purpose was not clear, and neither was how much damage the strike caused. But Isfahan is a major center of missile production, research and development for Iran, including the assembly of many of its Shahab medium-range missiles, which can reach Israel and beyond. Weeks ago, American officials publicly identified Iran as the primary supplier of drones to Russia for use in the war in Ukraine, and they said they believed Russia was also trying to obtain Iranian missiles to use in the conflict. But U.S. officials said they believed this strike was prompted by Israel's concerns about its own security, not the potential for missile exports to Russia. The strike came just as Secretary of State Antony J. Blinken was beginning a visit to Israel, his first since Benjamin Netanyahu returned to office as prime minister.