Alera

Abstract

The successful deployment of autonomous real-time systems is contingent on their ability to recover from performance degradation of sensors, actuators, and other electro-mechanical subsystems with low latency. In this article, we introduce ALERA, a novel framework for real-time control law adaptation in nonlinear control systems assisted by system state encodings that generate an error signal when the code properties are violated in the presence of failures. The fundamental contributions of this methodology are twofold-first, we show that the time-domain error signal contains perturbed system parameters' diagnostic information that can be used for quick control law adaptation to failure conditions and second, this quick adaptation is performed via reinforcement learning algorithms that relearn the control law of the perturbed system from a starting condition dictated by the diagnostic information, thus achieving significantly faster recovery. The fast (up to 80X faster than traditional reinforcement learning paradigms) performance recovery enabled by ALERA is demonstrated on an inverted pendulum balancing problem, a brake-by-wire system, and a self-balancing robot. CCS Concepts: • Computer systems organization → Maintainability and maintenance; Embedded and cyber-physical systems; • Computing methodologies → Markov decision processes;

A machine learning q‐RASPR approach for efficient predictions of the specific surface area of perovskites**

“…This can be observed in La 0.5 Bi 0.2 Ba 0.2 Mn 0.1 FeO 3 ( 38 ) which has a high CT value but a very low specific surface area value while in data points like LaMg 0.4 Cr 0.6 O 3 ( 22 ), the CT value is lower but the value of the specific surface area is much higher. Another RASPR descriptor Abs diff (| MaxPos‐MaxNeg |) signifies the absolute difference between the MaxPos and MaxNeg values [18, 24]. Since a greater number of compounds in our training set has a higher MaxPos value than the MaxNeg value, the descriptor Abs diff expresses a positive contribution to the response.…”

“…Read‐Across‐v4.1 available from https://sites.google.com/jadavpuruniversity.in/dtc-lab-software/home is a java‐based software tool that takes the training and test set files and a few hyperparameters as inputs and quickly computes the Read‐Across‐based predictions based on similarity considerations with the Euclidean Distance‐based approach, the Gaussian Kernel Similarity‐based approach and the Laplacian Kernel Similarity‐based approach. The tool also computes the corresponding validation metrics for predictions along with the compound‐specific similarity and error‐based measures for the confidence of predictions [15, 24]. The pre‐requisite to perform the Read‐Across‐based predictions is to identify the optimized setting of the hyperparameters.…”

Application of machine learning‐based read‐across structure‐property relationship (RASPR) as a new tool for predictive modelling: Prediction of power conversion efficiency (PCE) for selected classes of organic dyes in dye‐sensitized solar cells (DSSCs)

“…It is not a statistical method; instead, it uses the similarity between chemical analogues (source chemicals) to predict the property of unknown or query chemical substances [39]. This method uses either the analogue approach (one source chemical for the prediction) or the category approach (more than one source chemicals for predictions) [40]. The Read‐Across method is the most efficient alternative option for data gap‐filling and toxicity evaluation.…”

“…The RASPR method is a combined method of read‐across and QSPR that encapsulates the advantages of both of these methods and generates enhanced predictivity. In this method, selected structural and physicochemical descriptors from the QSPR model are used to generate different similarity and error‐based measures derived from the similarity‐based read‐across approach, [40] and these measures are merged with the initial structural and physicochemical descriptors to develop RASPR models after further feature selection. The general working process behind the RASPR is shown in Figure 1.…”