Search results

An invariants based architecture for combining small and large data sets in neural networks

We present a novel architecture for an AI system that allows a priori knowledge to combine with deep learning. In traditional neural networks, all available data is pooled at the input layer. Our alternative neural network is constructed so that partial representations (invariants) are learned in the intermediate layers, which can then be combined with a priori knowledge or with other predictive analyses of the same data. This leads to smaller training datasets due to more efficient learning. In addition, because this architecture allows inclusion of a priori knowledge and interpretable predictive models, the interpretability of the entire system increases while the data can still be used in a black box neural network. Our system makes use of networks of neurons rather than single neurons to enable the representation of approximations (invariants) of the output.

Social comparison in learning analytics dashboard supporting motivation and academic achievement

A promising contribution of Learning Analytics is the presentation of a learner's own learning behaviour and achievements via dashboards, often in comparison to peers, with the goal of improving self-regulated learning. However, there is a lack of empirical evidence on the impact of these dashboards and few designs are informed by theory. Many dashboard designs struggle to translate awareness of learning processes into actual self-regulated learning. In this study we investigate a Learning Analytics dashboard based on existing evidence on social comparison to support motivation, metacognition and academic achievement. Motivation plays a key role in whether learners will engage in self-regulated learning in the first place. Social comparison can be a significant driver in increasing motivation. We performed two randomised controlled interventions in different higher-education courses, one of which took place online due to the COVID-19 pandemic. Students were shown their current and predicted performance in a course alongside that of peers with similar goal grades. The sample of peers was selected in a way to elicit slight upward comparison. We found that the dashboard successfully promotes extrinsic motivation and leads to higher academic achievement, indicating an effect of dashboard exposure on learning behaviour, despite an absence of effects on metacognition. These results provide evidence that carefully designed social comparison, rooted in theory and empirical evidence, can be used to boost motivation and performance. Our dashboard is a successful example of how social comparison can be implemented in Learning Analytics Dashboards.

The Neural Network Zoo

From the article: Abstract: An overview of neural network architectures is presented. Some of these architectures have been created in recent years, whereas others originate from many decades ago. Apart from providing a practical tool for comparing deep learning models, the Neural Network Zoo also uncovers a taxonomy of network architectures, their chronology, and traces back lineages and inspirations for these neural information processing systems.

Deep learning to predict falls in older adults based on daily-life trunk accelerometry

The development and validation of a five factor model of sources of self-efficacy in clinical nursing education

Background: The aim of this study is to validate a newly developed nurses' self-efficacy sources inventory. We test the validity of a five-dimensional model of sources of self-efficacy, which we contrast with the traditional four-dimensional model based on Bandura's theoretical concepts. Methods: Confirmatory factor analysis was used in the development of the newly developed self-efficacy measure. Model fit was evaluated based upon commonly recommended goodness-of-fit indices, including the χ2 of the model fit, the Root Mean Square Error of approximation (RMSEA), the Tucker-Lewis Index (TLI), the Standardized Root Mean Square Residual (SRMR), and the Bayesian Information Criterion (BIC). Results: All 22 items of the newly developed five-factor sources of self-efficacy have high factor loadings (range .40-.80). Structural equation modeling showed that a five-factor model is favoured over the four-factor model. Conclusions and implications: Results of this study show that differentiation of the vicarious experience source into a peer- and expert based source reflects better how nursing students develop self-efficacy beliefs. This has implications for clinical learning environments: a better and differentiated use of self-efficacy sources can stimulate the professional development of nursing students.

An End-to-End Pipeline for Uncertainty Quantification and Remaining Useful Life Estimation: An Application on Aircraft Engines

Estimating the remaining useful life (RUL) of an asset lies at the heart of prognostics and health management (PHM) of many operations-critical industries such as aviation. Mod- ern methods of RUL estimation adopt techniques from deep learning (DL). However, most of these contemporary tech- niques deliver only single-point estimates for the RUL without reporting on the confidence of the prediction. This practice usually provides overly confident predictions that can have severe consequences in operational disruptions or even safety. To address this issue, we propose a technique for uncertainty quantification (UQ) based on Bayesian deep learning (BDL). The hyperparameters of the framework are tuned using a novel bi-objective Bayesian optimization method with objectives the predictive performance and predictive uncertainty. The method also integrates the data pre-processing steps into the hyperparameter optimization (HPO) stage, models the RUL as a Weibull distribution, and returns the survival curves of the monitored assets to allow informed decision-making. We vali- date this method on the widely used C-MAPSS dataset against a single-objective HPO baseline that aggregates the two ob- jectives through the harmonic mean (HM). We demonstrate the existence of trade-offs between the predictive performance and the predictive uncertainty and observe that the bi-objective HPO returns a larger number of hyperparameter configurations compared to the single-objective baseline. Furthermore, we see that with the proposed approach, it is possible to configure models for RUL estimation that exhibit better or comparable performance to the single-objective baseline when validated on the test sets.