В этом примере показано, как использовать Байесовую оптимизацию, чтобы выбрать оптимальные параметры для обучения классификатор ядра при помощи 'OptimizeHyperparameters'
аргумент значения имени. Выборочные данные установили airlinesmall.csv
большой набор данных, который содержит табличный файл данных о полете. Этот пример составляет длинную таблицу, содержащую данные, и извлекает метки класса и данные о предикторе из длинной таблицы, чтобы запустить процедуру оптимизации.
Когда вы выполняете вычисления на длинных массивах, MATLAB® использует любого параллельный пул (значение по умолчанию, если у вас есть Parallel Computing Toolbox™), или локальный сеанс работы с MATLAB. Если вы хотите запустить пример с помощью локального сеанса работы с MATLAB, когда у вас есть Parallel Computing Toolbox, можно изменить глобальную среду выполнения при помощи mapreducer
функция.
Создайте datastore, который ссылается на местоположение папки с данными. Данные могут содержаться в одном файле, наборе файлов или целой папке. Для папок, которые содержат набор файлов, можно задать целое местоположение папки или использовать подстановочный символ, '*.csv'
, включать несколько файлов с тем же расширением файла в datastore. Выберите подмножество переменных, чтобы работать с и обработать 'NA'
значения как недостающие данные так, чтобы datastore
заменяет их на NaN
значения. Составьте длинную таблицу, которая содержит данные в datastore.
ds = datastore('airlinesmall.csv'); ds.SelectedVariableNames = {'Month','DayofMonth','DayOfWeek',... 'DepTime','ArrDelay','Distance','DepDelay'}; ds.TreatAsMissing = 'NA'; tt = tall(ds) % Tall table
Starting parallel pool (parpool) using the 'local' profile ... Connected to the parallel pool (number of workers: 6). tt = M×7 tall table Month DayofMonth DayOfWeek DepTime ArrDelay Distance DepDelay _____ __________ _________ _______ ________ ________ ________ 10 21 3 642 8 308 12 10 26 1 1021 8 296 1 10 23 5 2055 21 480 20 10 23 5 1332 13 296 12 10 22 4 629 4 373 -1 10 28 3 1446 59 308 63 10 8 4 928 3 447 -2 10 10 6 859 11 954 -1 : : : : : : : : : : : : : :
Определите рейсы, которые являются поздними на 10 минут или больше путем определения логической переменной, которая верна для позднего рейса. Эта переменная содержит метки класса. Предварительный просмотр этой переменной включает первые несколько строк.
Y = tt.DepDelay > 10 % Class labels
Y = M×1 tall logical array 1 0 1 1 0 1 0 0 : :
Создайте длинный массив для данных о предикторе.
X = tt{:,1:end-1} % Predictor data
X = M×6 tall double matrix 10 21 3 642 8 308 10 26 1 1021 8 296 10 23 5 2055 21 480 10 23 5 1332 13 296 10 22 4 629 4 373 10 28 3 1446 59 308 10 8 4 928 3 447 10 10 6 859 11 954 : : : : : : : : : : : :
Удалите строки в X
и Y
это содержит недостающие данные.
R = rmmissing([X Y]); % Data with missing entries removed
X = R(:,1:end-1);
Y = R(:,end);
OptimizeHyperparameters
Оптимизируйте гиперпараметры автоматически с помощью 'OptimizeHyperparameters'
аргумент значения имени.
Стандартизируйте переменные предикторы.
Z = zscore(X);
Найдите оптимальные значения для 'KernelScale'
и 'Lambda'
аргументы значения имени, которые минимизируют потерю на наборе валидации затяжки. По умолчанию программное обеспечение выбирает и резервирует 20% данных как данные о валидации и обучает модель с помощью остальной части данных. Можно изменить часть затяжки при помощи 'HyperparameterOptimizationOptions'
аргумент значения имени. Для воспроизводимости используйте 'expected-improvement-plus'
функция приобретения и набор seed генераторов случайных чисел с помощью rng
и tallrng
. Результаты могут варьироваться в зависимости от количества рабочих и среды выполнения для длинных массивов. Для получения дополнительной информации смотрите Управление Где Ваши Запуски Кода.
rng('default') tallrng('default') Mdl = fitckernel(Z,Y,'Verbose',0,'OptimizeHyperparameters','auto',... 'HyperparameterOptimizationOptions',struct('AcquisitionFunctionName','expected-improvement-plus'))
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 2: Completed in 7.1 sec - Pass 2 of 2: Completed in 2.2 sec Evaluation completed in 12 sec
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.6 sec Evaluation completed in 1.8 sec |=====================================================================================================| | Iter | Eval | Objective | Objective | BestSoFar | BestSoFar | KernelScale | Lambda | | | result | | runtime | (observed) | (estim.) | | | |=====================================================================================================| | 1 | Best | 0.19672 | 125.49 | 0.19672 | 0.19672 | 1.2297 | 0.0080902 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.93 sec Evaluation completed in 1.1 sec | 2 | Accept | 0.19672 | 53.653 | 0.19672 | 0.19672 | 0.039643 | 2.5756e-05 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.5 sec Evaluation completed in 1.6 sec | 3 | Accept | 0.19672 | 52.453 | 0.19672 | 0.19672 | 0.02562 | 1.2555e-08 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 4 | Accept | 0.19672 | 57.223 | 0.19672 | 0.19672 | 92.644 | 1.2056e-07 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.4 sec Evaluation completed in 1.5 sec | 5 | Best | 0.11469 | 89.981 | 0.11469 | 0.12698 | 11.173 | 0.00024836 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.94 sec Evaluation completed in 1.1 sec | 6 | Best | 0.11365 | 82.031 | 0.11365 | 0.11373 | 10.609 | 0.00025761 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.92 sec Evaluation completed in 1.1 sec | 7 | Accept | 0.19672 | 50.604 | 0.11365 | 0.11373 | 0.0059498 | 0.00043861 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.89 sec Evaluation completed in 1 sec | 8 | Accept | 0.12122 | 91.341 | 0.11365 | 0.11371 | 11.44 | 0.00045722 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.97 sec Evaluation completed in 1.1 sec | 9 | Best | 0.10417 | 42.696 | 0.10417 | 0.10417 | 8.0424 | 6.7998e-05 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.87 sec Evaluation completed in 1 sec | 10 | Accept | 0.10433 | 42.215 | 0.10417 | 0.10417 | 9.6694 | 1.4948e-05 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.87 sec Evaluation completed in 1 sec | 11 | Best | 0.10409 | 41.618 | 0.10409 | 0.10411 | 6.2099 | 6.1093e-06 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.88 sec Evaluation completed in 1 sec | 12 | Best | 0.10383 | 44.635 | 0.10383 | 0.10404 | 5.6767 | 7.6134e-08 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.89 sec Evaluation completed in 1 sec | 13 | Accept | 0.10408 | 45.429 | 0.10383 | 0.10365 | 8.1769 | 8.5993e-09 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.89 sec Evaluation completed in 1 sec | 14 | Accept | 0.10404 | 41.928 | 0.10383 | 0.10361 | 7.6191 | 6.4079e-07 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.93 sec Evaluation completed in 1.1 sec | 15 | Best | 0.10351 | 42.094 | 0.10351 | 0.10362 | 4.2987 | 9.2645e-08 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.88 sec Evaluation completed in 1 sec | 16 | Accept | 0.10404 | 44.684 | 0.10351 | 0.10362 | 4.8747 | 1.7838e-08 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.87 sec Evaluation completed in 1 sec | 17 | Accept | 0.10657 | 88.006 | 0.10351 | 0.10357 | 4.8239 | 0.00016344 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.88 sec Evaluation completed in 1 sec | 18 | Best | 0.10299 | 41.303 | 0.10299 | 0.10358 | 3.5555 | 2.7165e-06 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.89 sec Evaluation completed in 1 sec | 19 | Accept | 0.10366 | 41.301 | 0.10299 | 0.10324 | 3.8035 | 1.3542e-06 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.87 sec Evaluation completed in 0.99 sec | 20 | Accept | 0.10337 | 41.345 | 0.10299 | 0.10323 | 3.806 | 1.8101e-06 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.89 sec Evaluation completed in 1 sec |=====================================================================================================| | Iter | Eval | Objective | Objective | BestSoFar | BestSoFar | KernelScale | Lambda | | | result | | runtime | (observed) | (estim.) | | | |=====================================================================================================| | 21 | Accept | 0.10345 | 41.418 | 0.10299 | 0.10322 | 3.3655 | 9.082e-09 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.86 sec Evaluation completed in 0.98 sec | 22 | Accept | 0.19672 | 60.129 | 0.10299 | 0.10322 | 999.62 | 1.2609e-06 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.88 sec Evaluation completed in 1 sec | 23 | Accept | 0.10315 | 41.133 | 0.10299 | 0.10306 | 3.6716 | 1.2445e-08 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.88 sec Evaluation completed in 1 sec | 24 | Accept | 0.19672 | 48.262 | 0.10299 | 0.10306 | 0.0010004 | 2.6214e-08 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.89 sec Evaluation completed in 1 sec | 25 | Accept | 0.19672 | 48.334 | 0.10299 | 0.10306 | 0.21865 | 0.0026529 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.86 sec Evaluation completed in 0.98 sec | 26 | Accept | 0.19672 | 60.229 | 0.10299 | 0.10306 | 299.92 | 0.0032109 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.87 sec Evaluation completed in 0.99 sec | 27 | Accept | 0.19672 | 48.361 | 0.10299 | 0.10306 | 0.002436 | 0.0040428 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.2 sec Evaluation completed in 1.4 sec | 28 | Accept | 0.19672 | 52.539 | 0.10299 | 0.10305 | 0.50559 | 3.3667e-08 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.88 sec Evaluation completed in 1 sec | 29 | Accept | 0.10354 | 43.957 | 0.10299 | 0.10313 | 3.7754 | 9.5626e-09 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 0.93 sec Evaluation completed in 1.1 sec | 30 | Accept | 0.10405 | 41.388 | 0.10299 | 0.10315 | 8.9864 | 2.3136e-07 |
__________________________________________________________ Optimization completed. MaxObjectiveEvaluations of 30 reached. Total function evaluations: 30 Total elapsed time: 1677.1387 seconds Total objective function evaluation time: 1645.7748 Best observed feasible point: KernelScale Lambda ___________ __________ 3.5555 2.7165e-06 Observed objective function value = 0.10299 Estimated objective function value = 0.10332 Function evaluation time = 41.3029 Best estimated feasible point (according to models): KernelScale Lambda ___________ __________ 3.6716 1.2445e-08 Estimated objective function value = 0.10315 Estimated function evaluation time = 42.3461
Mdl = ClassificationKernel PredictorNames: {'x1' 'x2' 'x3' 'x4' 'x5' 'x6'} ResponseName: 'Y' ClassNames: [0 1] Learner: 'svm' NumExpansionDimensions: 256 KernelScale: 3.6716 Lambda: 1.2445e-08 BoxConstraint: 665.9442 Properties, Methods
bayesopt
В качестве альтернативы можно использовать bayesopt
функционируйте, чтобы найти оптимальные значения гиперпараметров.
Разделите набор данных в наборы обучающих данных и наборы тестов. Задайте 1/3 выборку затяжки для набора тестов.
rng('default') % For reproducibility tallrng('default') % For reproducibility Partition = cvpartition(Y,'Holdout',1/3); trainingInds = training(Partition); % Indices for the training set testInds = test(Partition); % Indices for the test set
Извлеките данные об обучении и тестировании и стандартизируйте данные о предикторе.
Ytrain = Y(trainingInds); % Training class labels Xtrain = X(trainingInds,:); [Ztrain,mu,stddev] = zscore(Xtrain); % Standardized training data Ytest = Y(testInds); % Testing class labels Xtest = X(testInds,:); Ztest = (Xtest-mu)./stddev; % Standardized test data
Задайте переменные sigma
и lambda
найти оптимальные значения для 'KernelScale'
и 'Lambda'
аргументы значения имени. Используйте optimizableVariable
и задайте широкий спектр для переменных, потому что оптимальные значения неизвестны. Примените логарифмическое преобразование к переменным, чтобы искать оптимальные значения на логарифмической шкале.
N = gather(numel(Ytrain)); % Evaluate the length of the tall training array in memory
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: 0% complete Evaluation 0% complete
- Pass 1 of 1: Completed in 0.95 sec Evaluation 91% complete
Evaluation completed in 1.1 sec
sigma = optimizableVariable('sigma',[1e-3,1e3],'Transform','log'); lambda = optimizableVariable('lambda',[(1e-3)/N, (1e3)/N],'Transform','log');
Создайте целевую функцию для Байесовой оптимизации. Целевая функция берет в таблице, которая содержит переменные sigma
и lambda
, и затем вычисляет значение классификации потерь для бинарной Гауссовой модели классификации ядер, обученной с помощью fitckernel
функция. Установите 'Verbose',0
в fitckernel
подавить итеративное отображение диагностической информации.
minfn = @(z)gather(loss(fitckernel(Ztrain,Ytrain, ... 'KernelScale',z.sigma,'Lambda',z.lambda,'Verbose',0), ... Ztest,Ytest));
Оптимизируйте параметры [sigma,lambda]
из модели классификации ядер относительно потери классификации при помощи bayesopt
. По умолчанию, bayesopt
отображает итеративную информацию об оптимизации в командной строке. Для воспроизводимости установите AcquisitionFunctionName
опция к 'expected-improvement-plus'
. Функция приобретения по умолчанию зависит от времени выполнения и, поэтому, может дать различные результаты.
results = bayesopt(minfn,[sigma,lambda],'AcquisitionFunctionName','expected-improvement-plus')
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.3 sec |=====================================================================================================| | Iter | Eval | Objective | Objective | BestSoFar | BestSoFar | sigma | lambda | | | result | | runtime | (observed) | (estim.) | | | |=====================================================================================================| | 1 | Best | 0.19651 | 84.526 | 0.19651 | 0.19651 | 1.2297 | 0.012135 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 2 | Accept | 0.19651 | 112.57 | 0.19651 | 0.19651 | 0.039643 | 3.8633e-05 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.3 sec | 3 | Accept | 0.19651 | 80.282 | 0.19651 | 0.19651 | 0.02562 | 1.8832e-08 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.3 sec | 4 | Accept | 0.19651 | 52.306 | 0.19651 | 0.19651 | 92.644 | 1.8084e-07 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 5 | Accept | 0.19651 | 52.717 | 0.19651 | 0.19651 | 978.95 | 0.00015066 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 6 | Accept | 0.19651 | 90.336 | 0.19651 | 0.19651 | 0.0089609 | 0.0059189 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 7 | Accept | 0.19651 | 110.35 | 0.19651 | 0.19651 | 0.0010228 | 1.292e-08 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.3 sec | 8 | Accept | 0.19651 | 76.594 | 0.19651 | 0.19651 | 0.27475 | 0.0044831 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 9 | Accept | 0.19651 | 77.641 | 0.19651 | 0.19651 | 0.81326 | 1.0753e-07 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 10 | Accept | 0.19651 | 100.21 | 0.19651 | 0.19651 | 0.0040507 | 0.00011333 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 11 | Accept | 0.19651 | 52.287 | 0.19651 | 0.19651 | 964.67 | 1.2786e-08 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 12 | Accept | 0.19651 | 107.7 | 0.19651 | 0.19651 | 0.24069 | 0.0070503 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 13 | Accept | 0.19651 | 52.092 | 0.19651 | 0.19651 | 974.15 | 0.010898 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.3 sec | 14 | Accept | 0.19651 | 92.184 | 0.19651 | 0.19651 | 0.0013246 | 0.0011748 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 15 | Accept | 0.19651 | 87.893 | 0.19651 | 0.19651 | 0.0067415 | 1.9074e-07 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.3 sec | 16 | Accept | 0.19651 | 110.46 | 0.19651 | 0.19651 | 0.020448 | 1.247e-08 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 17 | Accept | 0.19651 | 104.12 | 0.19651 | 0.19651 | 0.0016556 | 0.0001784 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 18 | Accept | 0.19651 | 85.263 | 0.19651 | 0.19651 | 0.0047914 | 2.3289e-06 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 19 | Accept | 0.19651 | 52.102 | 0.19651 | 0.19651 | 90.015 | 0.00024412 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.3 sec | 20 | Accept | 0.19651 | 82.238 | 0.19651 | 0.19651 | 0.68775 | 2.7178e-07 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec |=====================================================================================================| | Iter | Eval | Objective | Objective | BestSoFar | BestSoFar | sigma | lambda | | | result | | runtime | (observed) | (estim.) | | | |=====================================================================================================| | 21 | Accept | 0.19651 | 49.468 | 0.19651 | 0.19651 | 49.073 | 0.00014766 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 22 | Accept | 0.19651 | 49.183 | 0.19651 | 0.19651 | 25.955 | 8.4946e-05 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 23 | Accept | 0.19651 | 84.781 | 0.19651 | 0.19651 | 0.002241 | 1.6284e-06 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 24 | Accept | 0.19651 | 90.023 | 0.19651 | 0.19651 | 0.060661 | 0.00041011 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 25 | Accept | 0.19651 | 87.349 | 0.19651 | 0.19651 | 0.035771 | 0.0023369 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.3 sec | 26 | Accept | 0.19651 | 49.932 | 0.19651 | 0.19651 | 713.45 | 3.5177e-08 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 27 | Accept | 0.19651 | 87.169 | 0.19651 | 0.19651 | 0.012395 | 1.8186e-06 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 28 | Accept | 0.19651 | 94.87 | 0.19651 | 0.19651 | 0.042872 | 0.0015886 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.2 sec | 29 | Best | 0.10795 | 37.932 | 0.10795 | 0.19346 | 1.5886 | 4.9128e-07 |
Evaluating tall expression using the Parallel Pool 'local': - Pass 1 of 1: Completed in 1.1 sec Evaluation completed in 1.3 sec | 30 | Accept | 0.19651 | 52.241 | 0.10795 | 0.19356 | 236.64 | 5.0506e-06 |
__________________________________________________________ Optimization completed. MaxObjectiveEvaluations of 30 reached. Total function evaluations: 30 Total elapsed time: 2455.5118 seconds Total objective function evaluation time: 2346.8025 Best observed feasible point: sigma lambda ______ __________ 1.5886 4.9128e-07 Observed objective function value = 0.10795 Estimated objective function value = 0.19356 Function evaluation time = 37.9317 Best estimated feasible point (according to models): sigma lambda ______ __________ 1.5886 4.9128e-07 Estimated objective function value = 0.19356 Estimated function evaluation time = 66.1901
results = BayesianOptimization with properties: ObjectiveFcn: @(z)gather(loss(fitckernel(Ztrain,Ytrain,'KernelScale',z.sigma,'Lambda',z.lambda,'Verbose',0),Ztest,Ytest)) VariableDescriptions: [1×2 optimizableVariable] Options: [1×1 struct] MinObjective: 0.1079 XAtMinObjective: [1×2 table] MinEstimatedObjective: 0.1936 XAtMinEstimatedObjective: [1×2 table] NumObjectiveEvaluations: 30 TotalElapsedTime: 2.4555e+03 NextPoint: [1×2 table] XTrace: [30×2 table] ObjectiveTrace: [30×1 double] ConstraintsTrace: [] UserDataTrace: {30×1 cell} ObjectiveEvaluationTimeTrace: [30×1 double] IterationTimeTrace: [30×1 double] ErrorTrace: [30×1 double] FeasibilityTrace: [30×1 logical] FeasibilityProbabilityTrace: [30×1 double] IndexOfMinimumTrace: [30×1 double] ObjectiveMinimumTrace: [30×1 double] EstimatedObjectiveMinimumTrace: [30×1 double]
Возвратите лучшую допустимую точку в модели Bayesian results
при помощи bestPoint
функция. Используйте критерий по умолчанию min-visited-upper-confidence-interval
, который определяет лучшую допустимую точку как посещаемую точку, которая минимизирует верхний доверительный интервал на значении целевой функции.
zbest = bestPoint(results)
zbest=1×2 table
sigma lambda
______ __________
1.5886 4.9128e-07
Таблица zbest
содержит оптимальные ориентировочные стоимости для 'KernelScale'
и 'Lambda'
аргументы значения имени. Можно задать эти значения когда обучение новый оптимизированный классификатор ядра при помощи
Mdl = fitckernel(Ztrain,Ytrain,'KernelScale',zbest.sigma,'Lambda',zbest.lambda)
Для длинных массивов может занять много времени процедура оптимизации. Если набор данных является слишком большим, чтобы запустить процедуру оптимизации, можно попытаться оптимизировать параметры только при помощи частичных данных. Используйте datasample
функционируйте и задайте 'Replace','false'
к выборочным данным без замены.
bayesopt
| bestPoint
| cvpartition
| datastore
| fitckernel
| gather
| loss
| optimizableVariable
| tall