biaohe Goto Github PK

artificial-intelligence

Include ML DL RL, knowledge and code

asynchronous-bo

Code related to the paper "Asynchronous Batch Bayesian Optimisation with Improved Local Penalisation"

autodiff

automatic differentiation made easier for C++

autograd

Efficiently computes derivatives of numpy code.

barefoot-framework

Batch Bayesian Optimization within the Model Reification and Fusion Framework

botorch

Bayesian optimization in PyTorch

casestudy

some spiceman for learning

continuousparetomtl

[ICML 2020] PyTorch Code for "Efficient Continuous Pareto Exploration in Multi-Task Learning"

den-armoea

# Introduction of DNN-AR-MOEA This repository contains code necessary to reproduce the experiments presented in Evolutionary Optimization of High-DimensionalMulti- and Many-Objective Expensive ProblemsAssisted by a Dropout Neural Network. Gaussian processes are widely used in surrogate-assisted evolutionary optimization of expensive problems. We propose a computationally efficient dropout neural network (EDN) to replace the Gaussian process and a new model management strategy to achieve a good balance between convergence and diversity for assisting evolutionary algorithms to solve high-dimensional multi- and many-objective expensive optimization problems. mainlydue to the ability to provide a confidence level of their outputs,making it possible to adopt principled surrogate managementmethods such as the acquisition function used in Bayesian opti-mization. Unfortunately, Gaussian processes become less practi-cal for high-dimensional multi- and many-objective optimizationas their computational complexity is cubic in the number oftraining samples. # References If you found DNN-AR-MOEA useful, we would be grateful if you cite the following reference: Evolutionary Optimization of High-DimensionalMulti- and Many-Objective Expensive ProblemsAssisted by a Dropout Neural Network (IEEE Transactions on Systems, Man and Cybernetics: Systems).

design-of-optimal-cmos-ring-oscillator-using-an-intelligent-optimization-tool

This paper presents an intelligent sizing method to improve the performance and efficiency of a CMOS Ring Oscillator (RO). The proposed approach is based on the simultaneous utilization of powerful and new multi-objective optimization techniques along with a circuit simulator under a data link. The proposed optimizing tool creates a perfect tradeoff between the contradictory objective functions in CMOS RO optimal design. This tool is applied for intelligent estimation of the circuit parameters (channel width of transistors), which have a decisive influence on RO specifications. Along the optimal RO design in an specified range of oscillaton frequency, the Power Consumption, Phase Noise, Figure of Merit (FoM), Integration Index, Design Cycle Time are considered as objective functions. Also, in generation of Pareto front some important issues, i.e. Overall Nondominated Vector Generation (ONVG), and Spacing (S) are considered for more effectiveness of the obtained feasible solutions in application. Four optimization algorithms called Multi-Objective Genetic Algorithm (MOGA), Multi-Objective Inclined Planes system Optimization (MOIPO), Multi-Objective Particle Swarm Optimization (MOPSO) and Multi-Objective Modified Inclined Planes System Optimization (MOMIPO) are utilized for 0.18-mm CMOS technology with supply voltage of 1-V. Baesd on our extensive simulations and experimental results MOMIPO outperforms the best performance among other multi-objective algorithms in presented RO designing tool.

dnn-mfbo

Multi-fidelity Bayesian Optimization via Deep Neural Nets

emukit

A Python-based toolbox of various methods in uncertainty quantification and statistical emulation: multi-fidelity, experimental design, Bayesian optimisation, Bayesian quadrature, etc.

enas

TensorFlow Code for paper "Efficient Neural Architecture Search via Parameter Sharing"

ensemble-bayesian-optimization

Code repository for Ensemble Bayesian Optimization

entropy-search

Entropy Search for Information-Efficient Global Optimization - JMLR v13

eshotgun

ϵ-shotgun: ϵ-greedy Batch Bayesian Optimisation

evolution-strategies-starter

Code for the paper "Evolution Strategies as a Scalable Alternative to Reinforcement Learning"

expected-uncertainty-reduction-for-kriging-based-reliability-analysis

In his project, we proposed a new acquisition function for kriging-based reliability analysis, namely expected uncertainty reduction (EUR), that serves as a meta-criterion to select the best sample from a set of optimal samples, each identified from a large number of candidate samples according to the criterion of an acquisition function.

ggp

Code and data for the paper `Bayesian Semi-supervised Learning with Graph Gaussian Processes'

gparml

Distributed Variational Inference in Sparse Gaussian Process Regression and Latent Variable Models.

gpss18

Gaussian Process and Uncertainty Quantification Summer School 2018

gpyopt

Gaussian Process Optimization using GPy

gpytorch

A highly efficient and modular implementation of Gaussian Processes in PyTorch

h2lib

H2Lib public repository

hetmogp

Heterogeneous Multi-output Gaussian Processes

hodlr

A fast, accurate direct solver and determinant computation for dense linear systems

icdm2016_b3o

Released code for ICDM 2016 Budgeted Batch Bayesian Optimization

ieci

Integrated Expected Conditional Improvement R implementation

incremental-gaussian-process-regression-igpr

jgan

JGAN model zoo supports 27 kinds of mainstream GAN models with high speed for jittor.