Dataset Information

A training algorithm for networks of high-variability reservoirs.

ABSTRACT: Physical reservoir computing approaches have gained increased attention in recent years due to their potential for low-energy high-performance computing. Despite recent successes, there are bounds to what one can achieve simply by making physical reservoirs larger. Therefore, we argue that a switch from single-reservoir computing to multi-reservoir and even deep physical reservoir computing is desirable. Given that error backpropagation cannot be used directly to train a large class of multi-reservoir systems, we propose an alternative framework that combines the power of backpropagation with the speed and simplicity of classic training algorithms. In this work we report our findings on a conducted experiment to evaluate the general feasibility of our approach. We train a network of 3 Echo State Networks to perform the well-known NARMA-10 task, where we use intermediate targets derived through backpropagation. Our results indicate that our proposed method is well-suited to train multi-reservoir systems in an efficient way.

SUBMITTER: Freiberger M

PROVIDER: S-EPMC7468150 | biostudies-literature | 2020 Sep

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

A training algorithm for networks of high-variability reservoirs.

Freiberger Matthias M Bienstman Peter P Dambre Joni J

Scientific reports 20200902 1

Physical reservoir computing approaches have gained increased attention in recent years due to their potential for low-energy high-performance computing. Despite recent successes, there are bounds to what one can achieve simply by making physical reservoirs larger. Therefore, we argue that a switch from single-reservoir computing to multi-reservoir and even deep physical reservoir computing is desirable. Given that error backpropagation cannot be used directly to train a large class of multi-res ...[more]

PMID: 32879360

Similar Datasets

Project description:Objective: In two independent study arms, we determine the effects of strength training (ST) and high-intensity interval training (HIIT) overload on cardiac autonomic modulation by measuring heart rate (HR) and vagal heart rate variability (HRV). Methods: In the study, 37 well-trained athletes (ST: 7 female, 12 male; HIIT: 9 female, 9 male) were subjected to orthostatic tests (HR and HRV recordings) each day during a 4-day baseline period, a 6-day overload microcycle, and a 4-day recovery period. Discipline-specific performance was assessed before and 1 and 4 days after training. Results: Following ST overload, supine HR, and vagal HRV (Ln RMSSD) were clearly increased and decreased (small effects), respectively, and the standing recordings remained unchanged. In contrast, HIIT overload resulted in decreased HR and increased Ln RMSSD in the standing position (small effects), whereas supine recordings remained unaltered. During the recovery period, these responses were reversed (ST: small effects, HIIT: trivial to small effects). The correlations between changes in HR, vagal HRV measures, and performance were weak or inconsistent. At the group and individual levels, moderate to strong negative correlations were found between HR and Ln RMSSD when analyzing changes between testing days (ST: supine and standing position, HIIT: standing position) and individual time series, respectively. Use of rolling 2-4-day averages enabled more precise estimation of mean changes with smaller confidence intervals compared to single-day values of HR or Ln RMSSD. However, the use of averaged values displayed unclear effects for evaluating associations between HR, vagal HRV measures, and performance changes, and have the potential to be detrimental for classification of individual short-term responses. Conclusion: Measures of HR and Ln RMSSD during an orthostatic test could reveal different autonomic responses following ST or HIIT which may not be discovered by supine or standing measures alone. However, these autonomic changes were not consistently related to short-term changes in performance and the use of rolling averages may alter these relationships differently on group and individual level.

Project description:BackgroundHigh talker variability (i.e., multiple voices in the input) has been found effective in training nonnative phonetic contrasts in adults. A small number of studies suggest that children also benefit from high-variability phonetic training with some evidence that they show greater learning (more plasticity) than adults given matched input, although results are mixed. However, no study has directly compared the effectiveness of high versus low talker variability in children.MethodsNative Greek-speaking eight-year-olds (N = 52), and adults (N = 41) were exposed to the English /i/-/ɪ/ contrast in 10 training sessions through a computerized word-learning game. Pre- and post-training tests examined discrimination of the contrast as well as lexical learning. Participants were randomly assigned to high (four talkers) or low (one talker) variability training conditions.ResultsBoth age groups improved during training, and both improved more while trained with a single talker. Results of a three-interval oddity discrimination test did not show the predicted benefit of high-variability training in either age group. Instead, children showed an effect in the reverse direction-i.e., reliably greater improvements in discrimination following single talker training, even for untrained generalization items, although the result is qualified by (accidental) differences between participant groups at pre-test. Adults showed a numeric advantage for high-variability but were inconsistent with respect to voice and word novelty. In addition, no effect of variability was found for lexical learning. There was no evidence of greater plasticity for phonetic learning in child learners.DiscussionThis paper adds to the handful of studies demonstrating that, like adults, child learners can improve their discrimination of a phonetic contrast via computerized training. There was no evidence of a benefit of training with multiple talkers, either for discrimination or word learning. The results also do not support the findings of greater plasticity in child learners found in a previous paper (Giannakopoulou, Uther & Ylinen, 2013a). We discuss these results in terms of various differences between training and test tasks used in the current work compared with previous literature.

Dataset Information

A training algorithm for networks of high-variability reservoirs.

Publications

A training algorithm for networks of high-variability reservoirs.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure