面试题与答案

Feature engineering involves transforming raw data into a format that is more suitable for modeling. It includes tasks like scaling, normalization, and creating new features to improve the performance of machine learning models.

The activation function introduces non-linearity to a neural network, allowing it to learn complex patterns. Common activation functions include sigmoid, tanh, and ReLU.

Precision is the ratio of correctly predicted positive observations to the total predicted positives. It is a measure of the accuracy of positive predictions made by a classification model.

Regularization is used to prevent overfitting in machine learning models by adding a penalty term to the cost function. It discourages the model from fitting the training data too closely and encourages generalization to new, unseen data.

A confusion matrix is a table used to evaluate the performance of a classification model. It compares the predicted and actual class labels, showing true positives, true negatives, false positives, and false negatives.

Hyperparameters are configuration settings for machine learning models that are not learned from the data but are set before the training process. Examples include learning rate, regularization strength, and the number of hidden layers in a neural network.

One-hot encoding is a technique used to represent categorical variables as binary vectors. Each category is represented by a unique binary value, with only one bit set to 1 and the rest set to 0. It is commonly used in machine learning algorithms that cannot work directly with categorical data.

A confusion matrix is a table that summarizes the performance of a classification algorithm. It shows the number of true positives, true negatives, false positives, and false negatives, providing insights into the model's accuracy, precision, recall, and other metrics.

Supervised learning involves training a model on a labeled dataset, while unsupervised learning deals with unlabeled data where the algorithm tries to find patterns or relationships on its own.

Cross-validation is a technique used to assess the performance of a model by dividing the dataset into multiple subsets, training the model on some, and testing on others. It helps to obtain a more reliable estimate of a model's performance.

Overfitting occurs when a model learns the training data too well, capturing noise and producing poor generalization on new data. Regularization techniques, cross-validation, and increasing training data are common methods to prevent overfitting.

A decision tree is a tree-like model where each node represents a decision based on a feature, and each branch represents an outcome of that decision. It is used for both classification and regression tasks.

Batch gradient descent updates the model parameters using the entire dataset, while stochastic gradient descent updates the parameters using one randomly selected data point at a time. Mini-batch gradient descent is a compromise, using a small subset of the data for each update.

KNN is a simple, instance-based learning algorithm used for classification and regression. It classifies a new data point based on the majority class of its k-nearest neighbors in the feature space.

The Receiver Operating Characteristic (ROC) curve is a graphical representation of a binary classification model's performance across different thresholds. It plots the true positive rate against the false positive rate, helping to assess the trade-off between sensitivity and specificity.

Dropout is a regularization technique used in neural networks to randomly deactivate some neurons during training. It helps prevent overfitting and encourages the network to learn more robust features.

Precision is the ratio of correctly predicted positive observations to the total predicted positives, while recall is the ratio of correctly predicted positive observations to the total actual positives. Precision emphasizes the accuracy of positive predictions, while recall focuses on capturing all positive instances.

Cross-entropy loss, or log loss, measures the performance of a classification model whose output is a probability value between 0 and 1. It penalizes models that are confidently wrong and is a common choice for binary and multiclass classification problems.

Precision is the ratio of true positives to the sum of true positives and false positives, while the F1 score is the harmonic mean of precision and recall. F1 score provides a balance between precision and recall, giving equal weight to both metrics.

Feature importance measures the contribution of each feature to the predictive performance of a model. It helps identify the most influential features in making predictions and is often used for feature selection and model interpretation.

Bias refers to the error introduced by approximating a real-world problem with a simplified model. Variance is the amount by which the model's prediction would change if it were estimated using a different training dataset. The bias-variance tradeoff aims to balance these two sources of error.

Ensemble learning combines the predictions of multiple models to improve overall performance. Common ensemble techniques include bagging, boosting, and stacking. The idea is that the combination of diverse models can provide better results than individual models.

The bias-variance tradeoff is a key concept in model selection. High bias leads to underfitting, while high variance leads to overfitting. It's about finding the right balance to achieve optimal model performance.

Bagging (Bootstrap Aggregating) and boosting are ensemble learning techniques. Bagging builds multiple models independently and combines them, while boosting builds models sequentially, giving more weight to misclassified instances.

The curse of dimensionality refers to the challenges and issues that arise when working with high-dimensional data. As the number of features increases, the data becomes sparse, and the computational requirements for training models grow exponentially.

L1 regularization adds the absolute values of the coefficients to the cost function, encouraging sparsity, while L2 regularization adds the squared values, penalizing large coefficients. L1 tends to produce sparse models, while L2 prevents extreme values in the coefficients.

Gradient boosting is an ensemble learning technique that builds a series of weak learners, typically decision trees, in a sequential manner. Each new learner corrects the errors of the previous ones, producing a strong, accurate model.

The learning rate determines the size of the steps taken during the optimization process. It is a hyperparameter that influences the convergence and stability of the optimization algorithm. A too-high learning rate may cause divergence, while a too-low rate may result in slow convergence.

Transfer learning is a technique where a pre-trained model on a large dataset is adapted for a different but related task. It allows leveraging knowledge gained from one domain to improve performance in another, often with smaller amounts of task-specific data.

Kernel functions in SVM enable the algorithm to operate in a higher-dimensional space without explicitly calculating the new feature space. They transform the input data into a higher-dimensional space, making it easier to find a hyperplane that separates different classes.