Showing papers on "Sampling (statistics) published in 2018"
TL;DR: This series of four articles intends to provide novice researchers with practical guidance for conducting high-quality qualitative research in primary care with references to criteria and tools for judging the quality of qualitative research papers.
Abstract: In the course of our supervisory work over the years, we have noticed that qualitative research tends to evoke a lot of questions and worries, so-called frequently asked questions (FAQs). This seri...
935 citations
Posted Content•
TL;DR: A single-layer recurrent neural network with a dual softmax layer that matches the quality of the state-of-the-art WaveNet model, the WaveRNN, and a new generation scheme based on subscaling that folds a long sequence into a batch of shorter sequences and allows one to generate multiple samples at once.
Abstract: Sequential models achieve state-of-the-art results in audio, visual and textual domains with respect to both estimating the data distribution and generating high-quality samples. Efficient sampling for this class of models has however remained an elusive problem. With a focus on text-to-speech synthesis, we describe a set of general techniques for reducing sampling time while maintaining high output quality. We first describe a single-layer recurrent neural network, the WaveRNN, with a dual softmax layer that matches the quality of the state-of-the-art WaveNet model. The compact form of the network makes it possible to generate 24kHz 16-bit audio 4x faster than real time on a GPU. Second, we apply a weight pruning technique to reduce the number of weights in the WaveRNN. We find that, for a constant number of parameters, large sparse networks perform better than small dense networks and this relationship holds for sparsity levels beyond 96%. The small number of weights in a Sparse WaveRNN makes it possible to sample high-fidelity audio on a mobile CPU in real time. Finally, we propose a new generation scheme based on subscaling that folds a long sequence into a batch of shorter sequences and allows one to generate multiple samples at once. The Subscale WaveRNN produces 16 samples per step without loss of quality and offers an orthogonal method for increasing sampling efficiency.
520 citations
TL;DR: LFADS, a deep learning method for analyzing neural population activity, can extract neural dynamics from single-trial recordings, stitch separate datasets into a single model, and infer perturbations, for example, from behavioral choices to these dynamics.
Abstract: Neuroscience is experiencing a revolution in which simultaneous recording of thousands of neurons is revealing population dynamics that are not apparent from single-neuron responses. This structure is typically extracted from data averaged across many trials, but deeper understanding requires studying phenomena detected in single trials, which is challenging due to incomplete sampling of the neural population, trial-to-trial variability, and fluctuations in action potential timing. We introduce latent factor analysis via dynamical systems, a deep learning method to infer latent dynamics from single-trial neural spiking data. When applied to a variety of macaque and human motor cortical datasets, latent factor analysis via dynamical systems accurately predicts observed behavioral variables, extracts precise firing rate estimates of neural dynamics on single trials, infers perturbations to those dynamics that correlate with behavioral choices, and combines data from non-overlapping recording sessions spanning months to improve inference of underlying dynamics.
455 citations
Posted Content•
TL;DR: This article proposed a probabilistic ensembles with trajectory sampling (PETS) algorithm, which combines uncertainty-aware deep network dynamics models with sampling-based uncertainty propagation to match the asymptotic performance of model-based and model-free deep RL algorithms.
Abstract: Model-based reinforcement learning (RL) algorithms can attain excellent sample efficiency, but often lag behind the best model-free algorithms in terms of asymptotic performance. This is especially true with high-capacity parametric function approximators, such as deep networks. In this paper, we study how to bridge this gap, by employing uncertainty-aware dynamics models. We propose a new algorithm called probabilistic ensembles with trajectory sampling (PETS) that combines uncertainty-aware deep network dynamics models with sampling-based uncertainty propagation. Our comparison to state-of-the-art model-based and model-free deep RL algorithms shows that our approach matches the asymptotic performance of model-free algorithms on several challenging benchmark tasks, while requiring significantly fewer samples (e.g., 8 and 125 times fewer samples than Soft Actor Critic and Proximal Policy Optimization respectively on the half-cheetah task).
391 citations
TL;DR: This article provides a very basic introduction to MCMC sampling, and describes what MCMC is, and what it can be used for, with simple illustrative examples.
Abstract: Markov Chain Monte–Carlo (MCMC) is an increasingly popular method for obtaining information about distributions, especially for estimating posterior distributions in Bayesian inference. This article provides a very basic introduction to MCMC sampling. It describes what MCMC is, and what it can be used for, with simple illustrative examples. Highlighted are some of the benefits and limitations of MCMC sampling, as well as different approaches to circumventing the limitations most likely to trouble cognitive scientists.
360 citations
Proceedings Article•
03 Jul 2018TL;DR: In this paper, a single-layer recurrent neural network (WaveRNN) with a dual softmax layer was proposed for text-to-speech synthesis, which achieved state-of-the-art results in audio, visual and textual domains.
Abstract: Sequential models achieve state-of-the-art results in audio, visual and textual domains with respect to both estimating the data distribution and generating high-quality samples. Efficient sampling for this class of models has however remained an elusive problem. With a focus on text-to-speech synthesis, we describe a set of general techniques for reducing sampling time while maintaining high output quality. We first describe a single-layer recurrent neural network, the WaveRNN, with a dual softmax layer that matches the quality of the state-of-the-art WaveNet model. The compact form of the network makes it possible to generate 24kHz 16-bit audio 4x faster than real time on a GPU. Second, we apply a weight pruning technique to reduce the number of weights in the WaveRNN. We find that, for a constant number of parameters, large sparse networks perform better than small dense networks and this relationship holds for sparsity levels beyond 96%. The small number of weights in a Sparse WaveRNN makes it possible to sample high-fidelity audio on a mobile CPU in real time. Finally, we propose a new generation scheme based on subscaling that folds a long sequence into a batch of shorter sequences and allows one to generate multiple samples at once. The Subscale WaveRNN produces 16 samples per step without loss of quality and offers an orthogonal method for increasing sampling efficiency.
358 citations
TL;DR: A novel approach that is based on Long Short-Term Memory (LSTM) is proposed that obtains higher accuracy in traffic flow prediction compared with other approaches.
310 citations
TL;DR: A survey on recent advances in distributed cooperative control under a sampled-data setting, with special emphasis on the published results since 2011, and several challenging issues for future research are proposed.
292 citations
21 May 2018
TL;DR: This paper proposes a methodology for nonuniform sampling, whereby a sampling distribution is learned from demonstrations, and then used to bias sampling, resulting in an order of magnitude improvement in terms of success rate and convergence to the optimal cost.
Abstract: A defining feature of sampling-based motion planning is the reliance on an implicit representation of the state space, which is enabled by a set of probing samples. Traditionally, these samples are drawn either probabilistically or deterministically to uniformly cover the state space. Yet, the motion of many robotic systems is often restricted to “small” regions of the state space, due to e.g. differential constraints or collision-avoidance constraints. To accelerate the planning process, it is thus desirable to devise non-uniform sampling strategies that favor sampling in those regions where an optimal solution might lie. This paper proposes a methodology for nonuniform sampling, whereby a sampling distribution is learned from demonstrations, and then used to bias sampling. The sampling distribution is computed through a conditional variational autoencoder, allowing sample generation from the latent space conditioned on the specific planning problem. This methodology is general, can be used in combination with any sampling-based planner, and can effectively exploit the underlying structure of a planning problem while maintaining the theoretical guarantees of sampling-based approaches. Specifically, on several planning problems, the proposed methodology is shown to effectively learn representations for the relevant regions of the state space, resulting in an order of magnitude improvement in terms of success rate and convergence to the optimal cost.
282 citations
TL;DR: This article categorizes, reviews, and analyzes the state-of-the-art single−/multi-response adaptive sampling approaches for global metamodeling in support of simulation-based engineering design and discusses some important issues that affect the success of an adaptive sampling approach.
Abstract: Metamodeling is becoming a rather popular means to approximate the expensive simulations in today’s complex engineering design problems since accurate metamodels can bring in a lot of benefits. The metamodel accuracy, however, heavily depends on the locations of the observed points. Adaptive sampling, as its name suggests, places more points in regions of interest by learning the information from previous data and metamodels. Consequently, compared to traditional space-filling sampling approaches, adaptive sampling has great potential to build more accurate metamodels with fewer points (simulations), thereby gaining increasing attention and interest by both practitioners and academicians in various fields. Noticing that there is a lack of reviews on adaptive sampling for global metamodeling in the literature, which is needed, this article categorizes, reviews, and analyzes the state-of-the-art single−/multi-response adaptive sampling approaches for global metamodeling in support of simulation-based engineering design. In addition, we also review and discuss some important issues that affect the success of an adaptive sampling approach as well as providing brief remarks on adaptive sampling for other purposes. Last, challenges and future research directions are provided and discussed.
276 citations
21 May 2018
TL;DR: This work presents PRM-RL, a hierarchical method for long-range navigation task completion that combines sampling-based path planning with reinforcement learning (RL), and evaluates it on two navigation tasks with non-trivial robot dynamics.
Abstract: We present PRM-RL, a hierarchical method for long-range navigation task completion that combines sampling-based path planning with reinforcement learning (RL). The RL agents learn short-range, point-to-point navigation policies that capture robot dynamics and task constraints without knowledge of the large-scale topology. Next, the sampling-based planners provide roadmaps which connect robot configurations that can be successfully navigated by the RL agent. The same RL agents are used to control the robot under the direction of the planning, enabling long-range navigation. We use the Probabilistic Roadmaps (PRMs) for the sampling-based planner. The RL agents are constructed using feature-based and deep neural net policies in continuous state and action spaces. We evaluate PRM-RL, both in simulation and on-robot, on two navigation tasks with non-trivial robot dynamics: end-to-end differential drive indoor navigation in office environments, and aerial cargo delivery in urban environments with load displacement constraints. Our results show improvement in task completion over both RL agents on their own and traditional sampling-based planners. In the indoor navigation task, PRM-RL successfully completes up to 215 m long trajectories under noisy sensor conditions, and the aerial cargo delivery completes flights over 1000 m without violating the task constraints in an environment 63 million times larger than used in training.
TL;DR: The usefulness and reliability of RAVE is demonstrated by applying it to model potentials of increasing complexity, including computation of the binding free energy profile for a hydrophobic ligand-substrate system in explicit water with dissociation time of more than 3 min in computer time at least twenty times less than that needed for umbrella sampling or metadynamics.
Abstract: Here we propose the reweighted autoencoded variational Bayes for enhanced sampling (RAVE) method, a new iterative scheme that uses the deep learning framework of variational autoencoders to enhance sampling in molecular simulations. RAVE involves iterations between molecular simulations and deep learning in order to produce an increasingly accurate probability distribution along a low-dimensional latent space that captures the key features of the molecular simulation trajectory. Using the Kullback-Leibler divergence between this latent space distribution and the distribution of various trial reaction coordinates sampled from the molecular simulation, RAVE determines an optimum, yet nonetheless physically interpretable, reaction coordinate and optimum probability distribution. Both then directly serve as the biasing protocol for a new biased simulation, which is once again fed into the deep learning module with appropriate weights accounting for the bias, the procedure continuing until estimates of desirable thermodynamic observables are converged. Unlike recent methods using deep learning for enhanced sampling purposes, RAVE stands out in that (a) it naturally produces a physically interpretable reaction coordinate, (b) is independent of existing enhanced sampling protocols to enhance the fluctuations along the latent space identified via deep learning, and (c) it provides the ability to easily filter out spurious solutions learned by the deep learning procedure. The usefulness and reliability of RAVE is demonstrated by applying it to model potentials of increasing complexity, including computation of the binding free energy profile for a hydrophobic ligand-substrate system in explicit water with dissociation time of more than 3 min, in computer time at least twenty times less than that needed for umbrella sampling or metadynamics.
08 Sep 2018
TL;DR: Spatiotemporal sampling network (STSN) as discussed by the authors performs object detection in a video frame by learning to spatially sample features from the adjacent frames, which is robust to occlusion or motion blur in individual frames.
Abstract: We propose a Spatiotemporal Sampling Network (STSN) that uses deformable convolutions across time for object detection in videos. Our STSN performs object detection in a video frame by learning to spatially sample features from the adjacent frames. This naturally renders the approach robust to occlusion or motion blur in individual frames. Our framework does not require additional supervision, as it optimizes sampling locations directly with respect to object detection performance. Our STSN outperforms the state-of-the-art on the ImageNet VID dataset and compared to prior video object detection methods it uses a simpler design, and does not require optical flow data for training.
21 May 2018
TL;DR: In this paper, the utility of Dropout Variational Inference (DVI) for object detection was investigated for the first time, and it was shown that label uncertainty can be extracted from a state-of-the-art object detection system via Dropout Sampling and used to increase object detection performance under the open-set conditions that are typically encountered in robotic vision.
Abstract: Dropout Variational Inference, or Dropout Sampling, has been recently proposed as an approximation technique for Bayesian Deep Learning and evaluated for image classification and regression tasks. This paper investigates the utility of Dropout Sampling for object detection for the first time. We demonstrate how label uncertainty can be extracted from a state-of-the-art object detection system via Dropout Sampling. We evaluate this approach on a large synthetic dataset of 30,000 images, and a real-world dataset captured by a mobile robot in a versatile campus environment. We show that this uncertainty can be utilized to increase object detection performance under the open-set conditions that are typically encountered in robotic vision. A Dropout Sampling network is shown to achieve a 12.3 % increase in recall (for the same precision score as a standard network) and a 15.1 % increase in precision (for the same recall score as the standard network).
Posted Content•
TL;DR: This work benchmarks well-established and recently developed methods for approximate posterior sampling combined with Thompson Sampling over a series of contextual bandit problems and finds that many approaches that have been successful in the supervised learning setting underperformed in the sequential decision-making scenario.
Abstract: Recent advances in deep reinforcement learning have made significant strides in performance on applications such as Go and Atari games. However, developing practical methods to balance exploration and exploitation in complex domains remains largely unsolved. Thompson Sampling and its extension to reinforcement learning provide an elegant approach to exploration that only requires access to posterior samples of the model. At the same time, advances in approximate Bayesian methods have made posterior approximation for flexible neural network models practical. Thus, it is attractive to consider approximate Bayesian neural networks in a Thompson Sampling framework. To understand the impact of using an approximate posterior on Thompson Sampling, we benchmark well-established and recently developed methods for approximate posterior sampling combined with Thompson Sampling over a series of contextual bandit problems. We found that many approaches that have been successful in the supervised learning setting underperformed in the sequential decision-making scenario. In particular, we highlight the challenge of adapting slowly converging uncertainty estimates to the online setting.
TL;DR: A new algorithm is proposed, TSEMO, which uses Gaussian processes as surrogates, which gives a simple algorithm without the requirement of a priori knowledge, reduced hypervolume calculations to approach linear scaling with respect to the number of objectives, the capacity to handle noise and the ability for batch-sequential usage.
Abstract: Many engineering problems require the optimization of expensive, black-box functions involving multiple conflicting criteria, such that commonly used methods like multiobjective genetic algorithms are inadequate. To tackle this problem several algorithms have been developed using surrogates. However, these often have disadvantages such as the requirement of a priori knowledge of the output functions or exponentially scaling computational cost with respect to the number of objectives. In this paper a new algorithm is proposed, TSEMO, which uses Gaussian processes as surrogates. The Gaussian processes are sampled using spectral sampling techniques to make use of Thompson sampling in conjunction with the hypervolume quality indicator and NSGA-II to choose a new evaluation point at each iteration. The reference point required for the hypervolume calculation is estimated within TSEMO. Further, a simple extension was proposed to carry out batch-sequential design. TSEMO was compared to ParEGO, an expected hypervolume implementation, and NSGA-II on nine test problems with a budget of 150 function evaluations. Overall, TSEMO shows promising performance, while giving a simple algorithm without the requirement of a priori knowledge, reduced hypervolume calculations to approach linear scaling with respect to the number of objectives, the capacity to handle noise and lastly the ability for batch-sequential usage.
TL;DR: In this paper, a comprehensive summary of existing measurement methods and evaluates their advantages, disadvantages, potential sources of error, and directions for future development is presented, as well as a comparison of direct and indirect methods.
TL;DR: An information-theoretic approach to stochastic optimal control problems that can be used to derive general sampling-based optimization schemes is presented and applied to the task of aggressive autonomous driving around a dirt test track.
Abstract: We present an information-theoretic approach to stochastic optimal control problems that can be used to derive general sampling-based optimization schemes. This new mathematical method is used to develop a sampling-based model predictive control algorithm. We apply this information-theoretic model predictive control scheme to the task of aggressive autonomous driving around a dirt test track, and compare its performance with a model predictive control version of the cross-entropy method.
TL;DR: It is shown that while current academic literature advocates probability sampling procedures, their actual usage is quite scarce and researchers should revisit the fundamental aspects of sampling to increase their research results’ rigour and relevance.
Abstract: In this research note, we reflect critically on the use of sampling techniques in advertising research. Our review of 1028 studies published between 2008 and 2016 in the four leading advertising journals shows that while current academic literature advocates probability sampling procedures, their actual usage is quite scarce. Most studies either lack information on the sampling method used, or engage in non-probability sampling without making adjustments to compensate for unequal selection probabilities, non-coverage, and sampling fluctuations. Based on our results, we call on researchers to revisit the fundamental aspects of sampling to increase their research results’ rigour and relevance.
TL;DR: Modelling how humans search for rewards under limited search horizons finds evidence that Gaussian process function learning—combined with an optimistic upper confidence bound sampling strategy—provides a robust account of how people use generalization to guide search.
Abstract: From foraging for food to learning complex games, many aspects of human behaviour can be framed as a search problem with a vast space of possible actions. Under finite search horizons, optimal solutions are generally unobtainable. Yet, how do humans navigate vast problem spaces, which require intelligent exploration of unobserved actions? Using various bandit tasks with up to 121 arms, we study how humans search for rewards under limited search horizons, in which the spatial correlation of rewards (in both generated and natural environments) provides traction for generalization. Across various different probabilistic and heuristic models, we find evidence that Gaussian process function learning—combined with an optimistic upper confidence bound sampling strategy—provides a robust account of how people use generalization to guide search. Our modelling results and parameter estimates are recoverable and can be used to simulate human-like performance, providing insights about human behaviour in complex environments.
Proceedings Article•
15 Feb 2018TL;DR: In this paper, approximate posterior sampling combined with Thompson sampling is used to balance exploration and exploitation in complex domains. But the authors highlight the challenge of adapting slowly converging uncertainty estimates to the online setting.
Abstract: Recent advances in deep reinforcement learning have made significant strides in performance on applications such as Go and Atari games. However, developing practical methods to balance exploration and exploitation in complex domains remains largely unsolved. Thompson Sampling and its extension to reinforcement learning provide an elegant approach to exploration that only requires access to posterior samples of the model. At the same time, advances in approximate Bayesian methods have made posterior approximation for flexible neural network models practical. Thus, it is attractive to consider approximate Bayesian neural networks in a Thompson Sampling framework. To understand the impact of using an approximate posterior on Thompson Sampling, we benchmark well-established and recently developed methods for approximate posterior sampling combined with Thompson Sampling over a series of contextual bandit problems. We found that many approaches that have been successful in the supervised learning setting underperformed in the sequential decision-making scenario. In particular, we highlight the challenge of adapting slowly converging uncertainty estimates to the online setting.
TL;DR: A stochastic extension of the primal-dual hybrid gradient algorithm studied by Chambolle and Pock in 2011 to solve saddle point problems that are separable in the dual variable is proposed.
Abstract: We propose a stochastic extension of the primal-dual hybrid gradient algorithm studied by Chambolle and Pock in 2011 to solve saddle point problems that are separable in the dual variable. The anal...
TL;DR: Online detection of false data injection attacks and denial of service attacks in the smart grid is studied and a novel event-based sampling scheme called level-crossing sampling with hysteresis is proposed that is shown to exhibit significant advantages compared with the conventional uniform-in-time sampling scheme.
Abstract: In this paper, online detection of false data injection attacks and denial of service attacks in the smart grid is studied. The system is modeled as a discrete-time linear dynamic system and state estimation is performed using the Kalman filter. The generalized cumulative sum algorithm is employed for quickest detection of the cyber-attacks. Detectors are proposed in both centralized and distributed settings. The proposed detectors are robust to time-varying states, attacks, and set of attacked meters. Online estimates of the unknown attack variables are provided, that can be crucial for a quick system recovery. In the distributed setting, due to bandwidth constraints, local centers can only transmit quantized messages to the global center, and a novel event-based sampling scheme called level-crossing sampling with hysteresis is proposed that is shown to exhibit significant advantages compared with the conventional uniform-in-time sampling scheme. Moreover, a distributed dynamic state estimator is proposed based on information filters. Numerical examples illustrate the fast and accurate response of the proposed detectors in detecting both structured and random attacks and their advantages over existing methods.
TL;DR: It is demonstrated that gradient designs outperform replicated designs for detecting and quantifying nonlinear responses, suggesting that a move to gradient designs in ecological experiments could be a major step towards unravelling underlying response patterns to continuous and interacting environmental drivers in a feasible and statistically powerful way.
Abstract: A fundamental challenge in experimental ecology is to capture nonlinearities of ecological responses to interacting environmental drivers. Here, we demonstrate that gradient designs outperform replicated designs for detecting and quantifying nonlinear responses. We report the results of (1) multiple computer simulations and (2) two purpose-designed empirical experiments. The findings consistently revealed that unreplicated sampling at a maximum number of sampling locations maximised prediction success (i.e. the R² to the known truth) irrespective of the amount of stochasticity and the underlying response surfaces, including combinations of two linear, unimodal or saturating drivers. For the two empirical experiments, the same pattern was found, with gradient designs outperforming replicated designs in revealing the response surfaces of underlying drivers. Our findings suggest that a move to gradient designs in ecological experiments could be a major step towards unravelling underlying response patterns to continuous and interacting environmental drivers in a feasible and statistically powerful way.
TL;DR: New exponential stabilization criteria dependent on and independent of upper bounds on time derivatives of fuzzy basis functions are established, by which a larger sampling interval can be achieved.
Abstract: This brief paper studies the exponential stabilization problem for the Takagi–Sugeno fuzzy systems with a variable sampling. Different from previous results, the gains of fuzzy state feedback controller adopted in this paper are time-varying during two consecutive sampling instants, which can contribute to the enlargement of the allowable sampling interval by choosing a suitable design parameter. To reduce the design conservativeness, a novel fuzzy time-dependent Lyapunov functional (FTDLF) is put forward to fully exploit the accessible information about the sampling pattern and the fuzzy basis functions. Moreover, a more relaxed constraint condition is presented to ensure the positive definiteness of the FTDLF on sampling intervals. By resorting to the novel FTDLF and the relaxed constraint condition, new exponential stabilization criteria dependent on and independent of upper bounds on time derivatives of fuzzy basis functions are established, by which a larger sampling interval can be achieved. One example is offered to demonstrate the validity and superiorities of the obtained new results.
TL;DR: A novel approach to automatically determine the locations for soil samples based on a soil map created from drone imaging after ploughing, and a wearable augmented reality technology to guide the user to the generated sample points is presented.
TL;DR: In this paper, the problem of sampling k-bandlimited signals on graphs is studied and two sampling strategies are proposed: one is non-adaptive and the other is adaptive but yields optimal results.
TL;DR: A review of 209 journal articles concerning OBIA published between 2003 and 2017 summarizes and discusses the current issues in object-based accuracy assessment to provide guidance for improved accuracy assessments for O BIA.
Abstract: Object-based image analysis (OBIA) has gained widespread popularity for creating maps from remotely sensed data. Researchers routinely claim that OBIA procedures outperform pixel-based procedures; however, it is not immediately obvious how to evaluate the degree to which an OBIA map compares to reference information in a manner that accounts for the fact that the OBIA map consists of objects that vary in size and shape. Our study reviews 209 journal articles concerning OBIA published between 2003 and 2017. We focus on the three stages of accuracy assessment: (1) sampling design, (2) response design and (3) accuracy analysis. First, we report the literature’s overall characteristics concerning OBIA accuracy assessment. Simple random sampling was the most used method among probability sampling strategies, slightly more than stratified sampling. Office interpreted remotely sensed data was the dominant reference source. The literature reported accuracies ranging from 42% to 96%, with an average of 85%. A third of the articles failed to give sufficient information concerning accuracy methodology such as sampling scheme and sample size. We found few studies that focused specifically on the accuracy of the segmentation. Second, we identify a recent increase of OBIA articles in using per-polygon approaches compared to per-pixel approaches for accuracy assessment. We clarify the impacts of the per-pixel versus the per-polygon approaches respectively on sampling, response design and accuracy analysis. Our review defines the technical and methodological needs in the current per-polygon approaches, such as polygon-based sampling, analysis of mixed polygons, matching of mapped with reference polygons and assessment of segmentation accuracy. Our review summarizes and discusses the current issues in object-based accuracy assessment to provide guidance for improved accuracy assessments for OBIA.
TL;DR: This work proposes a methodology to speed up the sampling of amorphous and disordered materials using a combination of a genetic algorithm and a specialized machine-learning potential based on artificial neural networks (ANNs).
Abstract: The atomistic modeling of amorphous materials requires structure sizes and sampling statistics that are challenging to achieve with first-principles methods. Here, we propose a methodology to speed up the sampling of amorphous and disordered materials using a combination of a genetic algorithm and a specialized machine-learning potential based on artificial neural networks (ANNs). We show for the example of the amorphous LiSi alloy that around 1000 first-principles calculations are sufficient for the ANN-potential assisted sampling of low-energy atomic configurations in the entire amorphous LixSi phase space. The obtained phase diagram is validated by comparison with the results from an extensive sampling of LixSi configurations using molecular dynamics simulations and a general ANN potential trained to ∼45 000 first-principles calculations. This demonstrates the utility of the approach for the first-principles modeling of amorphous materials.