The Fulu Foundation, a nonprofit that pays out bounties for removing user-hostile features, is hunting for a way to keep Ring cameras from sending data to Amazon--without breaking the hardware. Usually, when you see a feel-good story about finding a lost dog, you don't immediately react with fear and revulsion. But that was indeed the case in response to a Super Bowl commercial from Amazon-owned security camera company Ring. There's now a group offering to dole out a $10,000 bounty to wrest back control of the user data Ring controls. The ad showed off a new feature from Ring called Search Party.

Fulu sets repair bounties on consumer products that employ sneaky features that limit user control. Just this week, it awarded more than $10,000 to the person who hacked the Molekule air purifier. Companies tend to be rather picky about who gets to poke around inside their products. Manufacturers sometimes even take steps that prevent consumers from repairing their device when it breaks, or modifying it with third-party products. But those unsanctioned device modifications have become the raison d'être of a bounty program set up by a nonprofit called Fulu, or Freedom from Unethical Limitations on Users.

We propose a computationally efficient algorithm for gradient-based linear dimension reduction and high-dimensional regression. The algorithm initially computes a Mondrian forest and uses this estimator to identify a relevant feature subspace of the inputs from an estimate of the expected gradient outer product (EGOP) of the regression function. In addition, we introduce an iterative approach known as Transformed Iterative Mondrian (TrIM) forest to improve the Mondrian forest estimator by using the EGOP estimate to update the set of features and weights used by the Mondrian partitioning mechanism. We obtain consistency guarantees and convergence rates for the estimation of the EGOP matrix and the random forest estimator obtained from one iteration of the TrIM algorithm. Lastly, we demonstrate the effectiveness of our proposed algorithm for learning the relevant feature subspace across a variety of settings with both simulated and real data.

The acquisition and performance of arithmetic skills and basic operations such as addition, subtraction, multiplication, and division are essential for daily functioning, and reflect complex cognitive processes. This paper explores the cognitive mechanisms powering arithmetic learning, presenting a neurobiologically plausible cognitive architecture that simulates the acquisition of these skills. I implement a number vectorization embedding network and an associative memory model to investigate how an intelligent system can learn and recall arithmetic equations in a manner analogous to the human brain. I perform experiments that provide insights into the generalization capabilities of connectionist models, neurological causes of dyscalculia, and the influence of network architecture on cognitive performance. Through this interdisciplinary investigation, I aim to contribute to ongoing research into the neural correlates of mathematical cognition in intelligent systems.

Human learning is sensitive to rule-like structure and the curriculum of examples used for training. In tasks governed by succinct rules, learning is more robust when related examples are blocked across trials, but in the absence of such rules, interleaving is more effective. To date, no neural model has simultaneously captured these seemingly contradictory effects. Here we show that this same tradeoff spontaneously emerges with "in-context learning" (ICL) both in neural networks trained with metalearning and in large language models (LLMs). ICL is the ability to learn new tasks "in context" - without weight changes - via an inner-loop algorithm implemented in activation dynamics. Experiments with pretrained LLMs and metalearning transformers show that ICL exhibits the blocking advantage demonstrated in humans on a task involving rule-like structure, and conversely, that concurrent in-weight learning reproduces the interleaving advantage observed in humans on tasks lacking such structure.

"Today, the kind of superpowers that seem to belong in storybooks can be achieved by mathematical models, computation, new materials, AI, robotics–this convergence of fields," says Dr. Rus. This episode is part of our "Building the future" podcast series. It's a multi-episode series focusing on how organizations, researchers, and innovators are meeting our evolving global challenges. We understand the importance of inclusive conversations and have chosen to highlight the work of women on the cutting edge of technological innovation, and business excellence. A combination of technology and human ingenuity will push boundaries as companies look to enter a new wave of innovation through data and AI to enable growth. Although O'Reilly estimates that we're in the early stages of this transformation, she predicts that this convergence will be the biggest change since the industrial revolution.

I would love to see us change the fundamental paradigm for how we think about AI,

Early this week, tractor maker John Deere said it had signed a memorandum of understanding with the American Farm Bureau Federation, an agricultural trade group, promising to make it easier for farmers to access tools and software needed to repair their own equipment. The deal looked like a concession from the agricultural equipment maker, a major target of the right-to-repair movement, which campaigns for better access to documents and tools needed for people to repair their own gear. But right-to-repair advocates say that despite some good points, the agreement changes little, and farmers still face unfair barriers to maintaining equipment they own. Kevin O'Reilly, a director of the right-to-repair campaign run by the US Public Interest Research Group, a grassroots lobbying organization, says the timing of Deere's deal suggests the company may be trying to quash recent interest in right-to-repair laws from state legislators. In the past two years, corn belt states including Nebraska and Missouri, and also Montana, have considered giving farmers a legal right to tools needed to repair their own equipment. But no laws have been passed.

Understanding machine learning (ML) systems is a critical task for data scientists and non-technical profiles alike as organizations aim to integrate AI applications on an enterprise-wide level. In this ebook, we explore practices to identify cutting-edge and responsible strategies for managing high-impact AI systems and work to understand the concepts and techniques of model interpretability and explainability.

In the first quarter of 2022, global funding to artificial intelligence (AI) startups reached $15.1 billion, according to CB Insights' State of AI report. However, machine learning (ML) algorithms can lead to counterproductive results when deployed without reason. Here are four common challenges that companies implementing ML-based systems may encounter, along with some expert tips to maximize the impact of algorithms while avoiding missteps. For some companies, the first issues with AI and ML adoption come before starting. Machine learning is a vast, multifaceted discipline pervading most aspects of artificial intelligence.