Bayesian Deep Ensembles (BDEs) represent a powerful approach for uncertainty quantification in deep learning, combining the robustness of Deep Ensembles (DEs) with flexible multi-chain MCMC. While DEs are affordable in most deep learning settings, (long) sampling of Bayesian neural networks can be prohibitively costly. Yet, adding sampling after optimizing the DEs has been shown to yield significant improvements. This leaves a critical practical question: How long should the sequential sampling process continue to yield significant improvements over the initial optimized DE baseline? To tackle this question, we propose a stopping rule based on E-values. We formulate the ensemble construction as a sequential anytime-valid hypothesis test, providing a principled way to decide whether or not to reject the null hypothesis that MCMC offers no improvement over a strong baseline, to early stop the sampling. Empirically, we study this approach for diverse settings. Our results demonstrate the efficacy of our approach and reveal that only a fraction of the full-chain budget is often required.

America must win the AI energy war by 2026 or risk falling behind China, warns oil chief. Surging AI demand requires massive infrastructure buildup.

The entorhinal-hippocampal formation is the mammalian brain's navigation system, encoding both physical and abstract spaces via grid cells. This system is well-studied in neuroscience, and its efficiency and versatility make it attractive for applications in robotics and machine learning. While continuous attractor networks (CANs) successfully model entorhinal grid cells for encoding physical space, integrating both continuous spatial and abstract spatial computations into a unified framework remains challenging. Here, we attempt to bridge this gap by proposing a mechanistic model for versatile information processing in the entorhinal-hippocampal formation inspired by CANs and Vector Symbolic Architectures (VSAs), a neuro-symbolic computing framework. The novel grid-cell VSA (GC-VSA) model employs a spatially structured encoding scheme with 3D neuronal modules mimicking the discrete scales and orientations of grid cell modules, reproducing their characteristic hexagonal receptive fields. In experiments, the model demonstrates versatility in spatial and abstract tasks: (1) accurate path integration for tracking locations, (2) spatio-temporal representation for querying object locations and temporal relations, and (3) symbolic reasoning using family trees as a structured test case for hierarchical relationships.

Imagine it: you're on the bus or walking in the park, when you remember some important task has slipped your mind. You were meant to send an email, catch up on a meeting, or arrange to grab lunch with a friend. Without missing a beat, you simply say aloud what you've forgotten and the small device that's pinned to your chest, or resting on the bridge of your nose, sends the message, summarises the meeting, or pings your buddy a lunch invitation. The work has been taken care of, without you ever having to prod the screen of your smartphone. It's the sort of utopian convenience that a growing wave of tech companies are hoping to realise through artificial intelligence.

Systematic has developed a new AI-powered data analysis tool designed to transform data collection and management, providing commanders with an advanced data capture and analysis capability. SitaWare Insight decision support tool builds upon the successful SitaWare Headquarters C4I system. The new product enhances users' Intelligence Requirements Management (IRM) and Collection Management (CM) capabilities, making it easy to store and retrieve information from multiple sources. Henrik Sommer, SitaWare Insight product manager and domain expert, said, "SitaWare Insight exploits the'data lakes' that militaries control, formed from sensor information, images, videos, documents and more. The solution provides a secure, scalable repository for this information and an AI-powered search function, allowing users to pinpoint data on everything from enemy positions, images, documents to equipment. "It's like a military version of a commercial search engine, allowing access to the information you need, when you need it." It significantly enhances Intelligence Preparation of the Battlefield (IPB), offering advanced decision support through data collection, storage, and analysis. For example, operators could use SitaWare Insight to retrieve information on a particular geographical area, then combine this with the mapping capabilities of SitaWare Headquarters to significantly increase situational awareness and provide users with a range of useful data. Sommer points to Automatic Identification System (AIS) data in the naval domain, where SitaWare Insight can help identify patterns within such information: "SitaWare Insight AI can learn what is normal pattern of life in a certain area and then automatically detect and flag anomalies." SitaWare Insight is further enhanced through advanced image and object recognition technology. As such, it can be trained to identify different military objects like vehicles or equipment, tagging these with the appropriate metadata and storing them in the data lake for future access. "This means that later on, you can search for a specific vehicle type, a general topic, or information on a specific area," Sommer explained, adding that it also provides Natural Language Processing technology. "It can read a document of 100 pages or more then produce a one-page summary of key take aways." SitaWare Insight offers a scalable, future-proof solution through its underlying cloud-native architecture, which ensures security and the correct handling and management of data. It enables users to share data on a need-to-know basis and at different classification levels, defining specific user groups to manage access. SitaWare Insight is a significant enabler for the intelligence branch, operational planners, and the soldier on the ground, providing them with a holistic view of the battlefield. "Operational planners are fully dependant on information from the intelligence community.

Search engines scan the internet to find what you're looking for, or what you don't know you're looking for. Social media platforms surface content you might want to read. And the latest iPhones recognize your face in a split second to unlock your phone. As the artificial intelligence industry grows, the consequences of lacking a diverse workforce can pose major challenges that threaten to ripple into everyday life. Women and minorities are underrepresented in the AI workforce, according to a Stanford report on diversity in AI. More than 83% of AI tenure-track faculty at top universities are male, while over 46% of Ph.D. students in the United States studying AI are white.

Editor's Note: SHRM has partnered with TrainingIndustry.com to bring you relevant articles on key HR topics and strategies. The National Center for Women in Technology (NCWIT) predicts that while there will be 3.5 million "computing-related" jobs in the U.S. by 2026, 83% of them could go unfilled due to a lack of college graduates with related degrees. To meet this demand, organizations must reskill their workforces and look to candidates who have learned in-demand technical skills through alternate forms of education. In recent years, coding bootcamps have succeeded in training a diverse group of workers for careers as web, full-stack and software developers, among other roles, as well as reskilling people already in those professions. However, several major coding bootcamps have also closed in recent years, including Dev Bootcamp and The Iron Yard in 2017.

Another year has passed and humanity, for better or worse, remains in charge of the planet. Unfortunately for the robots, TNW has it on good authority they won't take over next year either. In the meantime, here's what the experts think will happen in 2019: Dialpad, an AI startup created by the original founders of Google Voice, tells TNW that all the hype over robot assistants that can make calls on your behalf may be a bit premature. Etienne Manderscheid, VP AI, Machine Learning, for the company says "robots may attempt to sound human next year, but this will work for few domains in 2019." Despite the hype brought on by Google Duplex and resulting conversations around speech synthesis, true text-to-speech technology will not be able to carry on conversations outside of the specific domains they're built around for at least another few years.

Most companies stick with their ERP system longer than people stay with their first spouse, which statistics estimate to be about eight years. Given the commitment involved, a move to an ERP cloud platform needs to be carefully planned, based not only on what your company needs today, but on a vision for the future. Brian Sommer, founder of technology advisory firm TechVentive in Carmel, Ind., pulls no punches when he talks about the problems with traditional ERP vendors' cloud offerings. He urges his clients to rethink their use of legacy ERP vendors because modern adaptations born in the cloud integrate AI and IoT capabilities and provide what businesses will need to compete in the years ahead. "Companies need to kick the tires on new vendors and think about how their competitors will use new technologies against their firm," he said.

Advances in artificial intelligence are changing the way automakers and their suppliers think about autonomous technology. Boosts in brainpower are not only accelerating the timeframe to bring self-driving technology to the marketplace, they're also broadening the scope of companies' ambition. "We're not just talking about autonomous cars," said Stefan Sommer, CEO of ZF Group. "We are talking about autonomous everything." At the CES technology show in Las Vegas, Sommer unveiled his company's newest product, an electronic control unit that contains artificial-intelligence software tailored for self-driving vehicles, including not only cars but also trains, buses, forklifts, trucks, tractors, and mining equipment.