Sample-Efficient Learning of Soft Task Priorities Through Bayesian Optimization

2018 ◽  
Author(s):  
Yinyin Su ◽  
Yuquan Wang ◽  
Abderrahmane Kheddar
Keyword(s):  
Bayesian Optimization ◽  
Efficient Learning

scholarly journals Task space adaptation via the learning of gait controllers of magnetic soft millirobots

2021 ◽  
pp. 027836492110218
Author(s):  
Sinan O. Demir ◽  
Utku Culha ◽  
Alp C. Karacakol ◽  
Abdon Pena-Francesch ◽  
Sebastian Trimpe ◽  
...  
Keyword(s):  
Human Body ◽  
Degrees Of Freedom ◽  
Bayesian Optimization ◽  
Small Scale ◽  
Soft Robots ◽  
Modeling And Control ◽  
Walking Gait ◽  
Physical Experiments ◽  
Efficient Learning ◽  
Task Environments

Untethered small-scale soft robots have promising applications in minimally invasive surgery, targeted drug delivery, and bioengineering applications as they can directly and non-invasively access confined and hard-to-reach spaces in the human body. For such potential biomedical applications, the adaptivity of the robot control is essential to ensure the continuity of the operations, as task environment conditions show dynamic variations that can alter the robot’s motion and task performance. The applicability of the conventional modeling and control methods is further limited for soft robots at the small-scale owing to their kinematics with virtually infinite degrees of freedom, inherent stochastic variability during fabrication, and changing dynamics during real-world interactions. To address the controller adaptation challenge to dynamically changing task environments, we propose using a probabilistic learning approach for a millimeter-scale magnetic walking soft robot using Bayesian optimization (BO) and Gaussian processes (GPs). Our approach provides a data-efficient learning scheme by finding the gait controller parameters while optimizing the stride length of the walking soft millirobot using a small number of physical experiments. To demonstrate the controller adaptation, we test the walking gait of the robot in task environments with different surface adhesion and roughness, and medium viscosity, which aims to represent the possible conditions for future robotic tasks inside the human body. We further utilize the transfer of the learned GP parameters among different task spaces and robots and compare their efficacy on the improvement of data-efficient controller learning.

scholarly journals Rapid Bayesian learning in the mammalian olfactory system

2019 ◽  
Author(s):  
Naoki Hiratani ◽  
Peter E. Latham
Keyword(s):  
Olfactory System ◽  
Bayesian Learning ◽  
Experimental Studies ◽  
Bayesian Optimization ◽  
Mammalian Brain ◽  
Theoretical Understanding ◽  
Plasticity Mechanism ◽  
Neural Architecture ◽  
Learning Rules ◽  
Efficient Learning

AbstractMany experimental studies suggest that animals can rapidly learn to identify odors and predict the rewards associated with them. However, the underlying plasticity mechanism remains elusive. In particular, it is not clear how olfactory circuits achieve rapid, data efficient learning with local synaptic plasticity. Here, we formulate olfactory learning as a Bayesian optimization process, then map the learning rules into a computational model of the mammalian olfactory circuit. The model is capable of odor identification from a small number of observations, while reproducing cellular plasticity commonly observed during development. We extend the framework to reward-based learning, and show that the circuit is able to rapidly learn odor-reward association with a plausible neural architecture. These results deepen our theoretical understanding of unsupervised learning in the mammalian brain.

Learning a Predictionless Resonating Controller for Wave Energy Converters

2019 ◽  
Author(s):  
Shuo Shi ◽  
Ron J. Patton ◽  
Mustafa Abdelrahman ◽  
Yanhua Liu
Keyword(s):  
Wave Energy ◽  
Bayesian Optimization ◽  
Irregular Waves ◽  
Complex Conjugate ◽  
Sea Waves ◽  
Sea State ◽  
Bayesian Optimization Algorithm ◽  
Physical Constraints ◽  
Model Free ◽  
Efficient Learning

Abstract This article presents a data-efficient learning approach for the complex-conjugate control of a wave energy point absorber. Particularly, the Bayesian Optimization algorithm is adopted for maximizing the extracted energy from sea waves subject to physical constraints. The algorithm learns the optimal coefficients of the causal controller. The simulation model of a Wavestar Wave Energy Converter (WEC) is selected to validate the control strategy for both the regular and irregular waves. The results indicate the efficiency and feasibility of the proposed control system. Less than 20 function evaluations are required to converge towards the optimal performance of each sea state. Additionally, this model-free controller can adapt to variations in the real sea state and be insensitive and robust to the WEC modeling bias.

Theoretical Studies of the Acid-Base Equilibria in a Model Active Site of the Human 20S Proteasome

2020 ◽  
Author(s):  
Jon Uranga ◽  
Lukas Hasecke ◽  
Jonny Proppe ◽  
Jan Fingerhut ◽  
Ricardo A. Mata
Keyword(s):  
Active Site ◽  
Boronic Acid ◽  
Computational Study ◽  
Ubiquitin Proteasome System ◽  
Bayesian Optimization ◽  
Inhibition Mechanism ◽  
20S Proteasome ◽  
Acid Base ◽  
Ubiquitin Proteasome ◽  
Infection Cycles

The 20S Proteasome is a macromolecule responsible for the chemical step in the ubiquitin-proteasome system of degrading unnecessary and unused proteins of the cell. It plays a central role both in the rapid growth of cancer cells as well as in viral infection cycles. Herein, we present a computational study of the acid-base equilibria in an active site of the human proteasome, an aspect which is often neglected despite the crucial role protons play in the catalysis. As example substrates, we take the inhibition by epoxy and boronic acid containing warheads. We have combined cluster quantum mechanical calculations, replica exchange molecular dynamics and Bayesian optimization of non-bonded potential terms in the inhibitors. In relation to the latter, we propose an easily scalable approach to the reevaluation of non-bonded potentials making use of QM/MM dynamics information. Our results show that coupled acid-base equilibria need to be considered when modeling the inhibition mechanism. The coupling between a neighboring lysine and the reacting threonine is not affected by the presence of the inhibitor.

Sign in / Sign up

Export Citation Format

Share Document