This paper uses chess, a landmark planning problem in AI, to assess transformers’ performance on a planning task where memorization is futile — even at a large scale. To this end, we release ChessBench, a large-scale benchmark dataset of 10 million chess games with legal move and value annotations (15 billion data points) provided by Stockfish 16, the state-of-the-art chess engine. We train transformers with up to 270 million parameters on ChessBench via supervised learning and perform extensive ablations to assess the impact of dataset size, model size, architecture type, and different prediction targets (state-values, action-values, and behavioral cloning). Our largest models learn to predict action-values for novel boards quite accurately, implying highly non-trivial generalization. Despite performing no explicit search, our resulting chess policy solves challenging chess puzzles and achieves a surprisingly strong Lichess blitz Elo of 2895 against humans (grandmaster level). We also compare to Leela Chess Zero and AlphaZero (trained without supervision via self-play) with and without search. We show that, although a remarkably good approximation of Stockfish’s search-based algorithm can be distilled into large-scale transformers via supervised learning, perfect distillation is still beyond reach, thus making ChessBench well-suited for future research.

There is strong interest in leveraging reinforcement learning (RL) for information retrieval (IR) applications including search, recommendation, and advertising. Just in 2020, the term "reinforcement learning" was mentioned in more than 60 different papers published by ACM SIGIR. It has also been reported that Internet companies like Google and Alibaba have started to gain competitive advantages from their RL-based search and recommendation engines. This full-day tutorial gives IR researchers and practitioners who have no or little experience with RL the opportunity to learn about the fundamentals of modern RL in a practical hands-on setting. Furthermore, some representative applications of RL in IR systems will be introduced and discussed. By attending this tutorial, the participants will acquire a good knowledge of modern RL concepts and standard algorithms such as REINFORCE and DQN. This knowledge will help them better understand some of the latest IR publications involving RL, as well as prepare them to tackle their own practical IR problems using RL techniques and tools. Please refer to the tutorial website (https://rl-starterpack.github.io/) for more information.

In this paper, we propose risk-calibrated evidential deep classifiers to reduce the costs associated with classification errors. We use two main approaches. The first is to develop methods to quantify the uncertainty of a classifier's predictions and reduce the likelihood of acting on erroneous predictions. The second is a novel way to train the classifier such that erroneous classifications are biased towards less risky categories. We combine these two approaches in a principled way. While doing this, we extend evidential deep learning with pignistic probabilities, which are used to quantify uncertainty of classification predictions and model rational decision making under uncertainty.We evaluate the performance of our approach on several image classification tasks. We demonstrate that our approach allows to (i) incorporate misclassification cost while training deep classifiers, (ii) accurately quantify the uncertainty of classification predictions, and (iii) simultaneously learn how to make classification decisions to minimize expected cost of classification errors.

The coarray programming model is an expression of the Single-Program-Multiple-Data (SPMD) programming model through the simple device of adding a codimension to the Fortran language. A data object declared with a codimension is a coarray object. Codimensions express the idea that some objects are located in local memory while others are located in remote memory. Coarray syntax obeys most of the same rules for normal array syntax. It is familiar to the Fortran programmer so the use of coarray syntax is natural and intuitive. Although the basic idea is quite simple, inserting it into the language definition turned out to be difficult. In addition, the process was complicated by rapidly changing hardware and heated arguments over whether parallelism should be supported best as an interface to language-independent libraries, as a set of directives superimposed on languages, or as a set of specific extensions to existing languages. In this paper, we review both the early history of coarrays and also their development into a part of Fortran 2008 and eventually into a larger part of Fortran 2018. Coarrays have been used, for example, in weather forecasting and in neural networks and deep learning.