Adversarial robustness of deep autoencoders (AEs) remains relatively unexplored, even though their non-invertible nature poses distinct challenges. Existing attack algorithms during the optimization of imperceptible, norm-bounded adversarial perturbations to maximize output damage in AEs, often stop at sub-optimal attacks. We observe that the adversarial loss gradient vanishes when backpropagated through ill-conditioned layers. This issue arises from near-zero singular values in the Jacobians of these layers, which weaken the gradient signal during optimization. We introduce GRILL, a technique that locally restores gradient signals in ill-conditioned layers, enabling more effective norm-bounded attacks. Through extensive experiments on different architectures of popular AEs, under both sample-specific and universal attack setups, and across standard and adaptive attack settings, we show that our method significantly increases the effectiveness of our adversarial attacks, enabling a more rigorous evaluation of AE robustness.

Forget smartwatches - in just five years time the next generation of Brit teenagers could be head-to-toe in AI-powered wearable tech, according to a new report. Scientist and leading futurist, Andrew Grill, claims that connected clothing will mean trainers will have the ability to'talk' to each other and collate style from other feet on the street. He believes that trainers could have a 5G-enabled chip built inside them, which will communicate to other people walking past. They'll hold data on your individual style, your preferences, budget and lifestyle, and collect insights on what new'looks' similar shoppers are trying out. Mr Grill told MailOnline: 'Because standalone 5G can connect more devices, it means that not only will it connect more people with phones, it can connect more things.

Generalization to unseen tasks is an important ability for few-shot learners to achieve better zero-/few-shot performance on diverse tasks. However, such generalization to vision-language tasks including grounding and generation tasks has been under-explored; existing few-shot VL models struggle to handle tasks that involve object grounding and multiple images such as visual commonsense reasoning or NLVR2. In this paper, we introduce GRILL, GRounded vIsion Language aLigning, a novel VL model that can be generalized to diverse tasks including visual question answering, captioning, and grounding tasks with no or very few training instances. Specifically, GRILL learns object grounding and localization by exploiting object-text alignments, which enables it to transfer to grounding tasks in a zero-/few-shot fashion. We evaluate our model on various zero-/few-shot VL tasks and show that it consistently surpasses the state-of-the-art few-shot methods.

Like a Roomba for your grill, this robot will help clean up after a barbecue. Futurism's mission is to empower our readers and drive the development of transformative technologies towards maximizing human potential.

"Flippy," a burger-grilling robot developed by Miso Robotics, is now operating at a Caliburger in Pasadena. After months of practice in the art of fast food preparation, "Flippy," has finally taken up a position as grill cook on the line at Caliburger's Pasadena, Calif., restaurant. "It's not a fun job -- it's hot, it's greasy, it's dirty," acknowledges John Miller, the CEO of Cali Group, which runs the international fast food chain. Even so, it could be the beginning of a bright career for Flippy in an industry that is otherwise notorious for high employee turnover. It certainly helps that Flippy, a burger-flipping robot developed by Miso Robotics, shows no concern about the low wages, meager benefits or long hours that plague the industry.