We present a new approach for identifying situations and behaviours, which we call moves, from soccer games in the 2D simulation league. Being able to identify key situations and behaviours are useful capabilities for analysing soccer matches, anticipating opponent behaviours to aid selection of appropriate tactics, and also as a prerequisite for automatic learning of behaviours and policies. To support a wide set of strategies, our goal is to identify situations from data, in an unsupervised way without making use of pre-defined soccer specific concepts such as "pass" or "dribble". The recurrent neural networks we use in our approach act as a high-dimensional projection of the recent history of a situation on the field. Similar situations, i.e., with similar histories, are found by clustering of network states. The same networks are also used to learn so-called conceptors, that are lower-dimensional manifolds that describe trajectories through a high-dimensional state space that enable situation-specific predictions from the same neural network. With the proposed approach, we can segment games into sequences of situations that are learnt in an unsupervised way, and learn conceptors that are useful for the prediction of the near future of the respective situation.

No abstract

Recurrent neural networks are a powerful means to cope with time series. We show how a type of linearly activated recurrent neural networks can approximate any timedependent function f (t) given by a number of function values. The approximation can effectively be learned by simply solving a linear equation system; no backpropagation or similar methods are needed. Furthermore, the network size can be reduced by taking only the most relevant components of the network. Thus, in contrast to others, our approach not only learns network weights but also the network architecture. The networks have interesting properties: They end up in ellipse trajectories in the long run and allow the prediction of further values and compact representations of functions. We demonstrate this by several experiments, among them multiple superimposed oscillators (MSO) and robotic soccer. Predictive neural networks outperform the previous state-of-the-art for the MSO task with a minimal number of units.

Introduction In 2020, the COVID-19 pandemic forced many schools to close their doors and transition to remote learning, disrupting how autistic students received school-based services and support. While school structure changes were challenging for all students, autistic students were uniquely affected, considering their reliance on predictability and routine; moreover, education settings are where most autistic children receive services. Much has been studied regarding the use of evidence-based practices (EBPs) for autistic students in traditional school settings, yet little is known about how educators use EBPs in remote learning environments in the wake of the COVID-19 pandemic. Method In this study, we explore educators’ experiences with EBP implementation at the height of the pandemic and educators’ reflections of its impact on autistic students and their school systems. Qualitative data were collected from 81 educators (general educators, special educators, and paraeducators) in semi-structured interviews regarding EBP use at the onset of the pandemic. Results Four themes emerged from interviews: (1) pandemic and remote learning environment challenges to inclusion and EBP use; (2) EBP use adaptations for remote learning environments; (3) pandemic and remote learning environment benefits for EBP use; and (4) considerations for EBP use beyond the pandemic. Conclusion These findings elucidate educators’ experiences using EBPs during the COVID-19 pandemic and highlight important areas of consideration for autism-focused EBP implementation as remote instruction continues to be a learning format. More research is needed to understand how to best implement EBPs for autistic students in this emerging instruction context.