At the end of season 2 of The Mandalorian (which aired in December 2020, so spoiler alerts are meaningless but here we are), a DeepFaked CGI version of Luke Skywalker makes a surprising appearance to recruit the young Grogu for his upcoming Jedi revival scheme. It was both a cool and creepy moment, with the impressive digital imaging nestling way too comfortably in the heart of the Uncanny Valley. It was so ineffably unsettling that a random YouTuber tried to fix the scene himself -- and did such a good job that Disney hired him. When CGI DeepFake Luke Skywalker returned in The Book of Boba Fett, his robotic performance was noticeably more convincing. But there was something about his voice that was still … off.

Out-of-distribution (OOD) detection is important for machine learning models deployed in the wild. Recent methods use auxiliary outlier data to regularize the model for improved OOD detection. However, these approaches make a strong distributional assumption that the auxiliary outlier data is completely separable from the in-distribution (ID) data. In this paper, we propose a novel framework that leverages wild mixture data -- that naturally consists of both ID and OOD samples. Such wild data is abundant and arises freely upon deploying a machine learning classifier in their \emph{natural habitats}. Our key idea is to formulate a constrained optimization problem and to show how to tractably solve it. Our learning objective maximizes the OOD detection rate, subject to constraints on the classification error of ID data and on the OOD error rate of ID examples. We extensively evaluate our approach on common OOD detection tasks and demonstrate superior performance.

St. Louis, Mo. (Ivanhoe Newswire) -- Artificial intelligence, or AI, allows machines to work more efficiently and solve problems faster. AI is all the buzz in the healthcare industry right now. And now, AI may also help to prevent some diseases. The same technology used in self-driving cars, smart assistants and disease mapping may also help to solve one of health care's biggest problems--how to stave off dementia. "What we're trying to do is intervene at that point when it starts to sharply decline to bring those skills back up," shared Adam Woods, PhD, University of Florida Center for Cognitive Aging and Memory.

Convolutional neural networks (CNNs) can now achieve human-level performance on challenging object recognition tasks. CNNs are also the leading quantitative models in terms of predicting neural and behavioral responses in visual recognition tasks. However, there is a widely accepted critique of CNN models: unlike newborn animals, which learn rapidly and efficiently, CNNs are thought to be "data hungry," requiring massive amounts of training data to develop accurate models for object recognition. This critique challenges the promise of using CNNs as models of visual development. Here, we directly examined whether CNNs are more data hungry than newborn animals by performing parallel controlled-rearing experiments on newborn chicks and CNNs. We raised newborn chicks in strictly controlled visual environments, then simulated the training data available in that environment by constructing a virtual animal chamber in a video game engine. We recorded the visual images acquired by an agent moving through the virtual chamber and used those images to train CNNs. When CNNs received similar visual training data as chicks, the CNNs successfully solved the same challenging view-invariant object recognition tasks as the chicks. Thus, the CNNs were not more data hungry than animals: both CNNs and chicks successfully developed robust object models from training data of a single object.

Not even your local pizza joint is safe from the forward progress of automation. At CES, a Seattle based Picnic showcased its automated pizza-making system that can swiftly assemble and cook pies with minimal human interaction. The system, which consists of three compact modular panels that assemble to form a conveyor belt, is capable of taking a pre-made pizza crust, adorning it with toppings, and cooking the pie to pre-specified doneness. What's even more compelling than the fact the pizza is made with little to no human input, however, is the speed at which Picnic's bot operates. According to CEO Clayton Wood, the bot can churn out an impressive 300 12-inch pizzas every hour when at max capacity.

The internet of things embeds intelligence into business processes to let us measure and manage the enterprise in ways that were never possible before. Amazon is enabling enterprises to bring your own algorithms to stream on its SageMaker machine learning service in a move designed to increase training speeds. Those algorithms will initially run on TensorFlow. Dr. Matt Wood, director of machine learning at AWS, said at the AWS Summit in New York that SageMaker is bringing streaming algorithms as well as batch job improvements to SageMaker. "We see a dramatic decrease in time to train models and put them into production," said Wood.

Machine learning was one of the main subjects on Tuesday (July 17) during the Amazon AWS Summit in New York City, as the eCommerce operator rolled out new cloud-based features related to commerce and payments. The announcements from Amazon Web Services (AWS) come at a time when machine learning and AI are receiving increased focus in the payments and commerce industries, with executives and researchers developing technologies that can operate outside the limitations of human bias and intuition. In fact, an upcoming PYMNTS webinar with Karen Webster and Sunil Madhu, founder of identity verification and fraud prevention services provider Socure, will include discussion about how a robust AI robot could leverage the full power of machine learning paired with massive data sets, drawing in data from online, offline and social sources. In New York on Tuesday, AWS said it had built machine learning and AI improvements into its cloud computing services designed for enterprises. Two of the improvements are meant to speed up the AWS SageMaker service, which customers can use to test and deploy custom AI algorithms.

NASA's Opportunity Mars rover has done many great things in its decade-plus of service--but initially, it rolled 600 feet past one of the initiative's biggest discoveries: the Block Island meteorite. Measuring about 67 centimeters across, the meteorite was a telltale sign that Mars' atmosphere had once been much thicker, thick enough to slow down the rock flying at a staggering 2km/s so that it did not disintegrate on impact. A thicker atmosphere could mean a more gentle climate, possibly capable of supporting liquid water on the surface, maybe even life. Yet, we only know about the Block Island meteorite because someone on the Opportunity science team manually spotted an unusual shape in low-resolution thumbnails of the images and decided it was worth backtracking for several days to examine it further. Instead of this machine purposefully heading toward the rock right from the get-go, the team barely saw perhaps its biggest triumph in the rear view mirror.

Addressing the South Australian government's recent Copper to the World conference in Adelaide, Newcrest's chief information and digital officer, Gavin Wood, gave a rundown on what had already been achieved at Newcrest with data science, virtual and augmented reality and artificial intelligence. He also talked about the benefits delivered by crowd sourcing, although this can also create some unique challenges of its own. "If you can imagine, an experienced operator at a site being told by a university student in Argentina the answer for optimising their part of the plant is quite different to something they believe from their experience of 20 or so years. Those are real challenges for our business," Wood said. He said data science coupled with machine learning had alr...

Amazon's DeepLens, a deep learning enabled video camera, is now generally available and hitting the market for $249. AWS DeepLens is designed to run models via TensorFlow and Caffe in less then 10 minute startup time for developers. The overall effort is to put more machine learning tool into the field and with developers. As for the hardware, DeepLens is a 4 megapixel camera with 1080P video, 2D microphone array, Intel Atom processor and 8GB of memory for models and code. The device runs Ubuntu 16.04, AWS Greengrass Core and optimized versions of MXNet and Intel clDNN libraries.