Algorithm selection can be performed using a model of runtime distribution, learned during a preliminary training phase. There is a trade-off between the performance of model-based algorithm selection, and the cost of learning the model. In this paper, we treat this trade-off in the context of bandit problems. We propose a fully dynamic and online algorithm selection technique, with no separate training phase: all candidate algorithms are run in parallel, while a model incrementally learns their runtime distributions. A redundant set of time allocators uses the partially trained model to propose machine time shares for the algorithms. A bandit problem solver mixes the model-based shares with a uniform share, gradually increasing the impact of the best time allocators as the model improves. We present experiments with a set of SAT solvers on a mixed SAT-UNSAT benchmark; and with a set of solvers for the Auction Winner Determination problem.Keywords algorithm selection · algorithm portfolios · online learning · life-long learning · bandit problem · expert advice · survival analysis · satisfiability · constraint programming Mathematics Subject Classifications (2000) 68T05 · 68T20 · 62N99 · 62G99

In recent years, gradient-based LSTM recurrent neural networks (RNNs) solved many previously RNN-unlearnable tasks. Sometimes, however, gradient information is of little use for training RNNs, due to numerous local minima. For such cases, we present a novel method: EVOlution of systems with LINear Outputs (Evolino). Evolino evolves weights to the nonlinear, hidden nodes of RNNs while computing optimal linear mappings from hidden state to output, using methods such as pseudoinverse-based linear regression. If we instead use quadratic programming to maximize the margin, we obtain the first evolutionary recurrent support vector machines. We show that Evolino-based LSTM can solve tasks that Echo State nets (Jaeger, 2004a) cannot and achieves higher accuracy in certain continuous function generation tasks than conventional gradient descent RNNs, including gradient-based LSTM.

The method of agent-based modeling is rarely used in social psychology, but has the potential to complement and improve traditional research practices. An agent-based model (ABM) consists of a number of virtual individuals-the "agents"-interacting in an artificial, experimenter-controlled environment. In this article, we discuss several characteristics of ABMs that could prove particularly useful with respect to recent recommendations aimed at countering issues related to the current "replication crisis". We address the potential synergies between planning and implementing an ABM on the one hand, and the endeavor of pre-registration on the other. We introduce ABMs as tools for both the generation and the improvement of theory, testing of hypotheses, and for extending traditional experimental approaches by facilitating the investigation of social processes from the intra-individual all the way up to the societal level. We describe examples of ABMs in social psychology, including a detailed description of the CollAct model of social learning. Finally, limitations and drawbacks of agent-based modeling are discussed. In annex 1 and 2, we provide literature and tool recommendations for getting started with an ABM.

Abstract. Given is a search problem or a sequence of search problems, as well as a set of potentially useful search algorithms. We propose a general framework for online allocation of computation time to search algorithms based on experience with their performance so far. In an example instantiation, we use simple linear extrapolation of performance for allocating time to various simultaneously running genetic algorithms characterized by different parameter values. Despite the large number of searchers tested in parallel, on various tasks this rather general approach compares favorably to a more specialized state-of-the-art heuristic; in one case it is nearly two orders of magnitude faster.