Bayesian optimization for goal-oriented multi-objective inverse material design.
Ontology highlight
ABSTRACT:
SUBMITTER: Hanaoka K
PROVIDER: S-EPMC8273421 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:In resource-constrained environments, such as low-power edge devices and smart sensors, deploying a fast, compact, and accurate intelligent system with minimum energy is indispensable. Embedding intelligence can be achieved using neural networks on neuromorphic hardware. Designing such networks would require determining several inherent hyperparameters. A key challenge is to find the optimum set of hyperparameters that might belong to the input/output encoding modules, the neural network itself, the application, or the underlying hardware. In this work, we present a hierarchical pseudo agent-based multi-objective Bayesian hyperparameter optimization framework (both software and hardware) that not only maximizes the performance of the network, but also minimizes the energy and area requirements of the corresponding neuromorphic hardware. We validate performance of our approach (in terms of accuracy and computation speed) on several control and classification applications on digital and mixed-signal (memristor-based) neural accelerators. We show that the optimum set of hyperparameters might drastically improve the performance of one application (i.e., 52-71% for Pole-Balance), while having minimum effect on another (i.e., 50-53% for RoboNav). In addition, we demonstrate resiliency of different input/output encoding, training neural network, or the underlying accelerator modules in a neuromorphic system to the changes of the hyperparameters.
| S-EPMC7396641 | biostudies-literature
Project description:Guiding experiments to find materials with targeted properties is a crucial aspect of materials discovery and design, and typically multiple properties, which often compete, are involved. In the case of two properties, new compounds are sought that will provide improvement to existing data points lying on the Pareto front (PF) in as few experiments or calculations as possible. Here we address this problem by using the concept and methods of optimal learning to determine their suitability and performance on three materials data sets; an experimental data set of over 100 shape memory alloys, a data set of 223 M2AX phases obtained from density functional theory calculations, and a computational data set of 704 piezoelectric compounds. We show that the Maximin and Centroid design strategies, based on value of information criteria, are more efficient in determining points on the PF from the data than random selection, pure exploitation of the surrogate model prediction or pure exploration by maximum uncertainty from the learning model. Although the datasets varied in size and source, the Maximin algorithm showed superior performance across all the data sets, particularly when the accuracy of the machine learning model fits were not high, emphasizing that the design appears to be quite forgiving of relatively poor surrogate models.
| S-EPMC5829239 | biostudies-literature
Project description:Excellent braking performance is the premise of safe driving, and improve the braking performance by upgrading structures and optimizing parameters of braking systems has become the pursuit of engineers. With the development of autonomous driving and intelligent connected vehicle, new structural schemes such as electro-mechanical brakes (EMBs) have become the future of vehicle braking systems. Meanwhile, many scholars have dedicated to the research on the parameters optimization of braking systems. While, most of the studies focus on reducing the brake size and weight, improving the brake responses by optimizing the parameters, almost not involving the braking performance, and the optimization variables are relatively single. On these foundations, a multi-objective optimal design of EMB parameters is proposed to enhance the vehicle's braking performance. Its objectives and constraints were defined based on relevant standards and regulations. Subsequently, the decision variables were set, and optimal math model was established. Furthermore, the co-simulation platform was constructed, and the optimal design and simulation analyses factoring in the crucial structural and control parameters were performed. The results confirmed that the maximum braking pressure response time of the EMB is decreased by approximately 0.3 s, the stopping distance (SD) of 90 km/h-0 is shortened by about 3.44 m. Moreover, the mean fully developed deceleration (MFDD) is increased by 0.002 g, and the lateral displacement of the body (LD) is reduced by about 0.037 m. Hence, the vehicle braking performance is improved.
| S-EPMC8133414 | biostudies-literature
Project description:Rapid engineering of biological systems is currently hindered by limited integration of manufacturing constraints into the design process, ultimately reducing the yield of many synthetic biology workflows. Here we tackle DNA engineering as a multi-objective optimization problem aiming at finding the best tradeoff between design requirements and manufacturing constraints. We developed a new open-source algorithm for DNA engineering, called Multi-Objective Optimisation algorithm for DNA Design and Assembly, available as a Python and Anaconda package, as well as a Docker image. Experimental results show that our method provides near-optimal constructs and scales linearly with design complexity, effectively paving the way to rational engineering of DNA molecules from genes to genomes.
| S-EPMC8524653 | biostudies-literature
Project description:BackgroundRNAs are attractive molecules as the biological parts for synthetic biology. In particular, the ability of conformational changes, which can be encoded in designer RNAs, enables us to create multistable molecular switches that function in biological circuits. Although various algorithms for designing such RNA switches have been proposed, the previous algorithms optimize the RNA sequences against the weighted sum of objective functions, where empirical weights among objective functions are used. In addition, an RNA design algorithm for multiple pseudoknot targets is currently not available.ResultsWe developed a novel computational tool for automatically designing RNA sequences which fold into multiple target secondary structures. Our algorithm designs RNA sequences based on multi-objective genetic algorithm, by which we can explore the RNA sequences having good objective function values without empirical weight parameters among the objective functions. Our algorithm has great flexibility by virtue of this weight-free nature. We benchmarked our multi-target RNA design algorithm with the datasets of two, three, and four target structures and found that our algorithm shows better or comparable design performances compared with the previous algorithms, RNAdesign and Frnakenstein. In addition to the benchmarks with pseudoknot-free datasets, we benchmarked MODENA with two-target pseudoknot datasets and found that MODENA can design the RNAs which have the target pseudoknotted secondary structures whose free energies are close to the lowest free energy. Moreover, we applied our algorithm to a ribozyme-based ON-switch which takes a ribozyme-inactive secondary structure when the theophylline aptamer structure is assumed.ConclusionsCurrently, MODENA is the only RNA design software which can be applied to multiple pseudoknot targets. Successful design results for the multiple targets and an RNA device indicate usefulness of our multi-objective RNA design algorithm.
| S-EPMC4559319 | biostudies-literature
Project description:Artificially synthesized RNA molecules have recently come under study since such molecules have a potential for creating a variety of novel functional molecules. When designing artificial RNA sequences, secondary structure should be taken into account since functions of noncoding RNAs strongly depend on their structure. RNA inverse folding is a methodology for computationally exploring the RNA sequences folding into a user-given target structure. In the present study, we developed a multi-objective genetic algorithm, MODENA (Multi-Objective DEsign of Nucleic Acids), for RNA inverse folding. MODENA explores the approximate set of weak Pareto optimal solutions in the objective function space of 2 objective functions, a structure stability score and structure similarity score. MODENA can simultaneously design multiple different RNA sequences at 1 run, whose lowest free energies range from a very stable value to a higher value near those of natural counterparts. MODENA and previous RNA inverse folding programs were benchmarked with 29 target structures taken from the Rfam database, and we found that MODENA can successfully design 23 RNA sequences folding into the target structures; this result is better than those of the other benchmarked RNA inverse folding programs. The multi-objective genetic algorithm gives a useful framework for a functional biomolecular design. Executable files of MODENA can be obtained at http://rna.eit.hirosaki-u.ac.jp/modena/.
| S-EPMC3169953 | biostudies-literature
Project description:Genetic algorithm is widely used in multi-objective mechanical structure optimization. In this paper, a genetic algorithm-based optimization method for ladle refractory lining structure is proposed. First, the parametric finite element model of the new ladle refractory lining is established by using ANSYS Workbench software. The refractory lining is mainly composed of insulating layer, permanent layer and working layer. Secondly, a mathematical model for multi-objective optimization is established to reveal the functional relationship between the maximum equivalent force on the ladle lining, the maximum temperature on the ladle shell, the total mass of the ladle and the structural parameters of the ladle refractory lining. Genetic algorithm translates the optimization process of ladle refractory lining into natural evolution and selection. The optimization results show that, compared with the unoptimized ladle refractory lining structure (insulation layer thickness of 0 mm, permanent layer thickness of 81 mm, and working layer thickness of 152 mm), the refractory lining with insulation layer thickness of 8.02 mm, permanent layer thickness of 76.20 mm, and working layer thickness of 148.61 mm has the best thermal insulation performance and longer service life within the variation of ladle refractory lining structure parameters. Finally, the results of the optimization are verified and analyzed in this paper. The study found that by optimizing the design of the ladle refractory lining, the maximum equivalent force on the ladle lining, the maximum temperature on the ladle shell and the ladle mass were reduced. The thermal insulation performance and the lightweight performance of the ladle are improved, which is very important for improving the service life of the ladle.
| S-EPMC9240744 | biostudies-literature
Project description:With the recent industrial expansion, heavy metals and other pollutants have increasingly contaminated our living surroundings. Heavy metals, being non-degradable, tend to accumulate in the food chain, resulting in potentially damaging toxicity to organisms. Thus, techniques to detect metal ions have gradually begun to receive attention. Recent progress in research on synthetic biology offers an alternative means for metal ion detection via the help of promoter elements derived from microorganisms. To make the design easier, it is necessary to develop a systemic design method for evaluating and selecting adequate components to achieve a desired detection performance. A multi-objective (MO) H2/H∞ performance criterion is derived here for design specifications of a metal ion biosensor to achieve the H2 optimal matching of a desired input/output (I/O) response and simultaneous H∞ optimal filtering of intrinsic parameter fluctuations and external cellular noise. According to the two design specifications, a Takagi-Sugeno (T-S) fuzzy model is employed to interpolate several local linear stochastic systems to approximate the nonlinear stochastic metal ion biosensor system so that the multi-objective H2/H∞ design of the metal ion biosensor can be solved by an associated linear matrix inequality (LMI)-constrained multi-objective (MO) design problem. The analysis and design of a metal ion biosensor with optimal I/O response matching and optimal noise filtering ability then can be achieved by solving the multi-objective problem under a set of LMIs. Moreover, a multi-objective evolutionary algorithm (MOEA)-based library search method is employed to find adequate components from corresponding libraries to solve LMI-constrained MO H2/H∞ design problems. It is a useful tool for the design of metal ion biosensors, particularly regarding the tradeoffs between the design factors under consideration.
| S-EPMC5104392 | biostudies-literature
Project description:Metallurgy and material design have thousands of years' history and have played a critical role in the civilization process of humankind. The traditional trial-and-error method has been unprecedentedly challenged in the modern era when the number of components and phases in novel alloys keeps increasing, with high-entropy alloys as the representative. New opportunities emerge for alloy design in the artificial intelligence era. Here, a successful machine-learning (ML) method has been developed to identify the microstructure images with eye-challenging morphology for a number of martensitic and ferritic steels. Assisted by it, a new neural-network method is proposed for the inverse design of alloys with 20 components, which can accelerate the design process based on microstructure. The method is also readily applied to other material systems given sufficient microstructure images. This work lays the foundation for inverse alloy design based on microstructure images with extremely similar features.
| S-EPMC8655171 | biostudies-literature
Project description:A new software tool making use of a genetic algorithm for multi-objective experimental optimization (GAME.opt) was developed based on a strength Pareto evolutionary algorithm. The software deals with high dimensional variable spaces and unknown interactions of design variables. This approach was evaluated by means of multi-objective test problems replacing the experimental results. A default parameter setting is proposed enabling users without expert knowledge to minimize the experimental effort (small population sizes and few generations).
| S-EPMC1705497 | biostudies-other