Today, deep neural networks are widely used since they can handle a variety of complex tasks. Their generality makes them very powerful tools in modern technology. However, deep neural networks are often overparameterized. The usage of these large models consumes a lot of computation resources. In this paper, we introduce a method called \textbf{T}ill the \textbf{L}ayers \textbf{C}ollapse (TLC), which compresses deep neural networks through the lenses of batch normalization layers. By reducing the depth of these networks, our method decreases deep neural networks' computational requirements and overall latency. We validate our method on popular models such as Swin-T, MobileNet-V2, and RoBERTa, across both image classification and natural language processing (NLP) tasks.

Ever since I watched the Disney Channel original movie Smart House as a kid, I've been fascinated by futuristic visions of technology in our daily lives. While writing about artificial intelligence tools and providing advice for using them at WIRED over the past two years, it's become so clear to me that the software is nothing like it's depicted in these sci-fi movies or what the hype-focused marketing materials from AI companies would have you believe. Even with this in mind, I do still consider the current crop of generative AI tools to be sometimes useful, sometimes entertaining, and almost always a little frustrating. And that belief was my driving motivation to write a second season of our AI Unlocked newsletter. It has been a passion project of mine over the past few months to work on this--chatting with experts in the field, trying out different tools, and soliciting reader responses to last year's newsletter.

Artificial intelligence is poised to deliver tremendous benefits to society. But, as many have pointed out, it could also bring unprecedented new horrors. As a general-purpose technology, the same tools that will advance scientific discovery could also be used to develop cyber, chemical, or biological weapons. Governing AI will require widely sharing its benefits while keeping the most powerful AI out of the hands of bad actors. The good news is that there is already a template on how to do just that.

While deep neural networks are highly effective at solving complex tasks, large pre-trained models are commonly employed even to solve consistently simpler downstream tasks, which do not necessarily require a large model's complexity. Motivated by the awareness of the ever-growing AI environmental impact, we propose an efficiency strategy that leverages prior knowledge transferred by large models. Simple but effective, we propose a method relying on an Entropy-bASed Importance mEtRic (EASIER) to reduce the depth of over-parametrized deep neural networks, which alleviates their computational burden. We assess the effectiveness of our method on traditional image classification setups. Our code is available at https://github.com/VGCQ/EASIER.

Since Google I/O in May, the company has heavily promoted its generative text and image AI tools to help people do everything from draft essays to create art. However, its core business model is selling ads and products. Today the company unveiled a new shopping tool that may help do exactly that. Now, customers in the United States can virtually "try on" women's tops. The company uses images of real models ranging from XXS to 3XL to wear AI-generated versions of clothes from hundreds of brands sold across Google, like Anthropologie, Everlane, and H&M.

The AI based tool can create "cutouts" or segments of different parts of an image. This comes handy while editing photos or while analyzing imagery for biological or security purposes. These tasks have one thing in common: you need to be able to identify and separate different objects within an image. Traditionally, researchers have had to start from scratch each time they want to analyze a new part of an image. Meta aims to change this laborious process by being the one-stop-shop for researchers and web developers working on such problems.

This paper investigates learning in a lifelong context. Lifelong learning addresses situations in which a learner faces a whole stream of learn(cid:173) ing tasks. Such scenarios provide the opportunity to transfer knowledge across multiple learning tasks, in order to generalize more accurately from less training data. In this paper, several different approaches to lifelong learning are described, and applied in an object recognition domain. It is shown that across the board, lifelong learning approaches generalize consistently more accurately from less training data, by their ability to transfer knowledge across learning tasks.

As the popular cliché has it, data scientists spend 80% of their time preparing the data, and only 20% developing the models.

Individuals and companies that suffer harm from drones, robots and other products or services equipped with artificial intelligence software will find it easier to sue for compensation under EU draft rules seen by Reuters. The AI Liability Directive, which the European Commission will announce on Wednesday, aims to address the increasing proliferation of AI-enabled products and services and the patchwork of national rules across the 27-country European Union. Victims can sue for compensation for harm to their life, property, health and privacy due to the fault or omission of a provider, developer or user of AI technology or was discriminated in a recruitment process using AI, the draft rules said. The rules seek to lighten the burden of proof on victims by introducing a "presumption of causality," which means victims only need to show that a manufacturer or user's failure to comply with certain requirements caused the harm and then link this to the AI technology in their lawsuit. Under a "right of access to evidence," victims can ask a court to order companies and suppliers to provide information about high-risk AI systems so that they can identify the liable person and find out what went wrong.

Carol Pinchefsky has written almost 2,000 articles about geek culture for outlets such as Forbes.com, Over the past 20 years, she's watched fantasy and science fiction grow from a niche interest to a massive cultural force. "There used to be a point where I knew everything there was to know about geek culture because it was contained within a few spheres," Pinchefsky says in Episode 504 of the Geek's Guide to the Galaxy podcast. I actually can't keep up. As a freelancer, Pinchefsky knows firsthand how hard it is for science fiction geeks to make a living doing what they love.