My next post will describe how to create deep-learning models that perform multiclass classification. In this post, you’ll learn how to use Keras to build binary classifiers. In fact, building a neural network that acts as a binary classifier is little different than building one that acts as a regressor. Such models are trained with datasets labeled with 1s and 0s representing the two classes, employ popular learning algorithms such as logistic regression and Naïve Bayes, and are frequently built with libraries such as Scikit-learn.ĭeep learning can be used for binary classification, too. Practical uses include sentiment analysis, spam detection, and credit-card fraud detection. Theįunction would need to take (y_true, y_pred) as arguments and returnĮither a single tensor value or a dict metric_name -> metric_value.One of the common uses for machine learning is performing binary classification, which looks at an input and predicts which of two possible classes it belongs to. Instead weighted towards penalizing incorrect class assignments.Ĭalculates the f-measure, the harmonic mean of precision and recall.Ĭustom metrics can be defined and passed via the compilation step. With beta = 1, this is equivalent to a F-measure. the proportion of incorrect class assignments. The F-beta score (ranged from 0.0 to 1.0)Ĭomputes this, as a weighted mean of the proportion of correct classĪssignments vs. In order to avoid this, a metric should penalize incorrect classĪssignments as well (recall). Would achieve a perfect score by simply assigning every class to every By only using accuracy (precision) a model This is useful for multi-label classification, where input samples can beĬlassified as sets of labels. Matthews_correlation matthews_correlation(y_true, y_pred)Ĭalculates the Matthews correlation coefficient measure for qualityĬalculates the precision, a metric for multi-label classification ofĬalculates the recall, a metric for multi-label classification ofįbeta_score fbeta_score(y_true, y_pred, beta=1)Ĭalculates the F score, the weighted harmonic mean of precision and recall. Kullback_leibler_divergence kullback_leibler_divergence(y_true, y_pred)Ĭalculates the Kullback-Leibler (KL) divergence between predictionĬalculates the poisson function over prediction and target values.Ĭosine_proximity cosine_proximity(y_true, y_pred)Ĭalculates the cosine similarity between the prediction and target If you get a shape error, add a length-1 dimensionīinary_crossentropy binary_crossentropy(y_true, y_pred)Ĭalculates the cross-entropy value for binary classification Labels shape must have the same number of dimensions as
Note: Expects an array of integerĬlasses. Sparse_categorical_crossentropy sparse_categorical_crossentropy(y_true, y_pred) Note: Expects a binary class matrix instead of a vector Squared_hinge squared_hinge(y_true, y_pred)Ĭalculates the squared value of the hinge loss.Ĭategorical_crossentropy categorical_crossentropy(y_true, y_pred)Ĭalculates the cross-entropy value for multiclass classification Mean_squared_logarithmic_error mean_squared_logarithmic_error(y_true, y_pred)Ĭalculates the mean squared logarithmic error (msle) rateĬalculates the hinge loss, which is defined as Mean_absolute_percentage_error mean_absolute_percentage_error(y_true, y_pred)Ĭalculates the mean absolute percentage error (mape) rate
Mean_absolute_error mean_absolute_error(y_true, y_pred)Ĭalculates the mean absolute error (mae) rate Mean_squared_error mean_squared_error(y_true, y_pred)Ĭalculates the mean squared error (mse) rate Target class is within the top-k predictions provided. Top_k_categorical_accuracy top_k_categorical_accuracy(y_true, y_pred, k=5)Ĭalculates the top-k categorical accuracy rate, i.e. Same as categorical_accuracy, but useful when the predictions are for Sparse_categorical_accuracy sparse_categorical_accuracy(y_true, y_pred)
Single tensor value representing the mean of the output array across allĪvailable metrics binary_accuracy binary_accuracy(y_true, y_pred)Ĭalculates the mean accuracy rate across all predictions for binaryĬategorical_accuracy categorical_accuracy(y_true, y_pred)Ĭalculates the mean accuracy rate across all predictions for Theano/TensorFlow tensor of the same shape as y_true. You can either pass the name of an existing metric, or pass a Theano/TensorFlow symbolic function (see Custom metrics). Metric functions are to be supplied in the metrics parameter when a model is compiled.Ī metric function is similar to an objective function, except that the results from evaluating a metric are not used when training the model. A metric is a function that is used to judge the performance of your model.
0 kommentar(er)
