Russian forces have been using drones to hunt and attack civilians in Ukraine and continue to do so, according to Human Rights Watch (HRW). In a report released on Tuesday, HRW stated that the Russian military has repeatedly deployed unmanned drones to attack civilian targets in its more than three-year war with Ukraine. The NGO said that dozens of civilians have been killed and hundreds injured in violation of the laws of war. Referencing video from Russian drones and witnesses and survivors, the rights watchdog alleges that Russia has "deliberately or recklessly" hunted civilians and civilian objects, particularly in the southern city of Kherson, using "commercially available quadcopter drones" made domestically and in China. "Russian drone operators are able to track their targets, with high-resolution video feeds, leaving little doubt that the intent is to kill, maim, and terrify civilians," Belkis Wille, a director on arms and conflict at HRW, said in a statement.

Ukraine's surprise Sunday attack on Russian offensive weapons caches may be a good time for the U.S. to reflect on its own weaknesses, should one of its adversaries attempt a similar strike. Col. Seth Krummrich, a retired Army Special Forces commander and vice president at the Virginia-based security firm Global Guardian, warned that the U.S. remains vulnerable to drone attacks. "Interestingly, it is not a technological gap, it is a policy/authority process to engage and deny drone attacks," Krummrich said. "I assess it will take a major drone strike in the U.S. to change policy." Even civilian operations have a tough time getting approval for drone-interception-authority protections, the NFL excepted, he said.

What message does Ukraine's Operation Spider's Web send to Russia and US? Ukraine carries out large-scale drone strikes on multiple Russian airbases.Read more Eighteen months in the making, Ukraine's Operation Spider's Web saw hundreds of AI-trained drones target military aircraft deep inside Russia's borders. Ukrainian President Volodymyr Zelenskyy says Sunday's attacks will go down in history. He followed them up with a proposal for an unconditional ceasefire as the two sides met in Istanbul. The European Union is preparing its 18th package of sanctions on Russia, while US President Donald Trump has threatened to use "devastating" measures against Russia if he feels the time is right. So, is the time right now?

Lanyard-clad attendees with branded tote bags and pink-shirted volunteers flowed through London's Brick Lane on Monday, marking the launch of the inaugural SXSW London festival. Taking place over multiple stages and venues in Shoreditch and Hoxton, SXSW London has officially kicked off its first full day of panels, keynotes, demonstrations, movie premieres, and music gigs. And luckily, Londoners are no strangers to a queue, with SXSW's penchant for long lines outside Austin venues replicated in the UK capital. Playing to the strengths of fellow conferences, the biggest topics of SXSW London are the impact of AI on essentially anything you could think of, the creator economy and online communities, and self-driving tech -- I spied a Wayve autonomous vehicle carefully navigating the pedestrian-filled Brick Lane (with a human driver behind the wheel, just in case). London mayor Sadiq Khan officially launched the festival with a speech Monday morning, championing London as "a global centre for AI investment and innovation," emphasising a focus on ethical and accessible AI development, and playing to the audience with a ChatGPT anecdote.

Decoding non-invasive brain recordings is pivotal for advancing our understanding of human cognition but faces challenges due to individual differences and complex neural signal representations. Traditional methods often require customized models and extensive trials, lacking interpretability in visual reconstruction tasks.

Graphical models are powerful tools for modeling high-dimensional data, but learning graphical models in the presence of latent variables is well-known to be difficult. In this work we give new results for learning Restricted Boltzmann Machines, probably the most well-studied class of latent variable models.

Training and inference with large machine learning models that far exceed the memory capacity of individual devices necessitates the design of distributed architectures, forcing one to contend with communication constraints. We present a framework for distributed computation over a quantum network in which data is encoded into specialized quantum states. We prove that for models within this framework, inference and training using gradient descent can be performed with exponentially less communication compared to their classical analogs, and with relatively modest overhead relative to standard gradient-based methods. We show that certain graph neural networks are particularly amenable to implementation within this framework, and moreover present empirical evidence that they perform well on standard benchmarks. To our knowledge, this is the first example of exponential quantum advantage for a generic class of machine learning problems that hold regardless of the data encoding cost. Moreover, we show that models in this class can encode highly nonlinear features of their inputs, and their expressivity increases exponentially with model depth. We also delineate the space of models for which exponential communication advantages hold by showing that they cannot hold for linear classification. Communication of quantum states that potentially limit the amount of information that can be extracted from them about the data and model parameters may also lead to improved privacy guarantees for distributed computation. Taken as a whole, these findings form a promising foundation for distributed machine learning over quantum networks.

In this paper we provide a general framework for designing, analyzing and implementing such algorithms in the episodic reinforcement learning problem. This framework is built upon Lagrangian duality, and demonstrates that every model-optimistic algorithm that constructs an optimistic MDP has an equivalent representation as a value-optimistic dynamic programming algorithm. Typically, it was thought that these two classes of algorithms were distinct, with model-optimistic algorithms benefiting from a cleaner probabilistic analysis while value-optimistic algorithms are easier to implement and thus more practical. With the framework developed in this paper, we show that it is possible to get the best of both worlds by providing a class of algorithms which have a computationally efficient dynamic-programming implementation and also a simple probabilistic analysis. Besides being able to capture many existing algorithms in the tabular setting, our framework can also address large-scale problems under realizable function approximation, where it enables a simple model-based analysis of some recently proposed methods.

Data curation is an essential component of large-scale pretraining. In this work, we demonstrate that jointly prioritizing batches of data is more effective for learning than selecting examples independently. Multimodal contrastive objectives expose the dependencies between data and thus naturally yield criteria for measuring the joint learnability of a batch. We derive a simple and tractable algorithm for selecting such batches, which significantly accelerate training beyond individuallyprioritized data points. As performance improves by selecting from large superbatches, we also leverage recent advances in model approximation to reduce the computational overhead of scoring.

This paper investigates simple bilevel optimization problems where we minimize an upper-level objective over the optimal solution set of a convex lower-level objective. Existing methods for such problems either only guarantee asymptotic convergence, have slow sublinear rates, or require strong assumptions. To address these challenges, we propose a penalization framework that delineates the relationship between approximate solutions of the original problem and its reformulated counterparts.