The different roles of c and P in this lower bound inspire us to design algorithms that estimate costs and transitions separately. Specifically, assuming the knowledge of c and P, we develop a simple but sub-optimal algorithm and another more involved minimax optimal algorithm (up to logarithmic terms). These algorithms combine the ideas of finite-horizon approximation [Chen et al., 2022a], special Bernstein-style bonuses of the MVP algorithm[Zhangetal.,2020],adaptiveconfidencewidening[WeiandLuo,2021],as well as some new techniques such as properly penalizing long-horizon policies. Finally,when c and P are unknown, we develop avariant ofthe MASTER algorithm [Weiand Luo,2021]and integrate the aforementioned ideas into itto achieve O(min{B?S

We study the Stochastic Shortest Path (SSP) problem in which an agent has to reach a goal state in minimum total expected cost. In the learning formulation ofthe problem, the agent has no prior knowledge about the costs and dynamics of the model. She repeatedly interacts with the model forK episodes, and has to minimize her regret. In this work we show that the minimax regret for this setting is eO( p (B2?+B?)|S||A|K)whereB?

Automated slice classification is clinically relevant since it can be incorporated into medical image segmentation workflows as a preprocessing step that would flag slices with a higher probability of containing tumors, thereby directing physicians' attention to the important slices. In this work, we train a ResNet-18 network to classify axial slices of lymphoma PET/CT images (collected from two institutions) depending on whether the slice intercepted a tumor (positive slice) in the 3D image or if the slice did not (negative slice). Various instances of the network were trained on 2D axial datasets created in different ways: (i) slice-level split and (ii) patient-level split; inputs of different types were used: (i) only PET slices and (ii) concatenated PET and CT slices; and different training strategies were employed: (i) center-aware (CAW) and (ii) center-agnostic (CAG). Model performances were compared using the area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPRC), and various binary classification metrics. We observe and describe a performance overestimation in the case of slice-level split as compared to the patient-level split training. The model trained using patient-level split data with the network input containing only PET slices in the CAG training regime was the best performing/generalizing model on a majority of metrics. Our models were additionally more closely compared using the sensitivity metric on the positive slices from their respective test sets.

Whether it's robots, computers or genetically engineered beings, technology is out to get humanity. It's trying to enslave or kill us, or make us suffer on behalf of some corrupt corporate or government entity. But is it merely fanciful fiction or the bleak reality that awaits mankind? Many of our favourite science-fiction films are part of a genre known as "tech noir", stories that prophesy that the advancement of technology will have foreboding consequences for humanity. Blockbusters such as The Terminator franchise, Blade Runner, RoboCop and Gattaca dissolve the distinction between man and machine, exploring what it means to be human and challenging contemporary values, says Dr Greg Dolgopolov, lecturer in the school of media, film and theatre at the University of NSW.

TELLING Hiroshi Ishiguro apart from his robotic replica can be tough. The professor's silicon-brainchild, cast from plastic and plaster moulds of his face and body and implanted with some of his sweeping black locks, is as realistic as a Madame Tussaud's waxworks model - down to the frown on his broad forehead. The android, Geminoid H1-1, mimics its master's posture, lip movements and facial gestures - even his fidgety fingers and feet. It blinks, seems to breathe and sounds, for all the world, like its human counterpart. Remote-controlled Geminoid stands in when the busy professor doesn't have time to lecture and presents a compelling scripted talk.

IN 1951, a man invented a fabric that never gets dirty and doesn't wear out. He should have been lauded but instead he was threatened and hounded by those who feared he was about to kill the clothing industry with his one suit that would last a lifetime. The good news is that he was a work of fiction, played by Alec Guinness in the British Ealing Studios comedy The Man in the White Suit. But believe it or not, exactly such a fabric is not far away. The modern version could result in a suit that can house computers, recharge batteries and even put on a light show.

The perfect game of draughts ends as a draw, Canadian computer scientists reported on Thursday. The team at the University of Alberta said they had "solved" draughts, the 5000-year-old popular board game also known as chequers (or checkers). Their computer program, Chinook, spent more than 18 years playing out the 500 billion possible positions, they report in the journal Science. "This paper announces that checkers is now solved: Perfect play by both sides leads to a draw," Jonathan Schaeffer and colleagues wrote in their report. "That checkers is a draw is not a surprise; grandmaster players have conjectured this for decades."

But the robotic rush in Japan is also being driven by unique societal needs. Confronting a major depopulation problem due to a record low birthrate and its status as the nation with the longest lifespan on Earth, Japanese are fretting about who will staff the factory floors of the world's second-largest economy in the years ahead. Toyota, Japan's biggest car maker, has come up with one answer in moving to create a line of worker robots with human-like hands able to perform multiple sophisticated tasks.

Rodney Brooks, the Australian-born head of the artificial intelligence lab at the Massachusetts Institute of Technology, believes an affordable child-replacement is a long way off. "It is only Japanese companies that are pushing humanoid robots as commercial products and I do not yet see the cost-effectiveness of them; certainly not for the next 10 years, and maybe not for at least 50 years."