What is the employment prospect of artificial intelligence now ? Do you want to find a job after completing courses related to artificial intelligence ? At present, the field of artificial intelligence is developing rapidly in China , Major enterprises are actively laying out the application of artificial intelligence technology . Want to work on artificial intelligence , We should not only understand the overall development direction of China . Only the foundation is the best stepping stone to enter the enterprise . I found some interview questions about artificial intelligence in the dark horse programmer forum today . I hope it will be of some help to the small partners seeking knowledge in the near future .
Artificial intelligence interview question sharing （ With answers ）
1, Deep learning framework TensorFlow What are the four commonly used cross entropy ?
answer ： tf.nn.weighted_cross_entropy_with_logits
tf.nn.sigmoid_cross_entropy_with_logits
tf.nn.softmax_cross_entropy_with_logits
tf.nn.sparse_softmax_cross_entropy_with_logits
2, What is over fitting , What are the measures to avoid over fitting ?
answer ： Over fitting ： In machine learning , When we test the model , Improved expressiveness in training data sets ,
But the performance in the training set decreased .
Solution ：
1. Regularization ;
2. In the process of training the model , tune parameter . The learning rate should not be too high ;
3. Cross validation of data ;
4. Select the percentage of test set data suitable for the training set , Select the appropriate stop training standard , Train the machine properly ;
5. In the neural network model , We can reduce the weight ;
3, What is kernel function ?
Kernel function is to project linearly indivisible features into high-order feature space , So that the support vector machine can be linearly separable in this high-dimensional space , That is, the kernel function can map to high-dimensional space and solve nonlinear classification problems . Including linear kernel function , Polynomial kernel function , Gaussian kernel function, etc , Gaussian kernel function is the most commonly used .
4, Deep learning framework TensorFlow What are the common kernel functions in ?
SVM Gaussian kernel function , Should be if you want to split a non-linear dataset , Change the linear classifier to insinuate to the data set , It's about to change SVM Kernel function in loss function
Linear kernel function
Polynomial kernel function
5, What are the advantages of naive Bayesian method ?
Naive Bayes has stable classification efficiency
It performs well for small-scale data , Able to deal with multi classification problems , When the data exceeds memory , De incremental training
Less sensitive to missing data , The algorithm is relatively simple , Commonly used for text classification .
6, What is the standard method of supervised learning ?
All regression algorithms and classification algorithms belong to supervised learning, and the initial values are given clearly. There are features and labels in the training set , And a model is obtained through training , When facing data with only features but no labels , Can predict .
7, In machine learning , What is the choice of model ?
According to a set of models with different complexity , Select the best model from a model . After selecting the best model , Evaluate the prediction error and other evaluation indicators on the new data .
8, Graphic database Neo4J Advantages and disadvantages of ?
advantage ：1. Faster database operations , The premise is that the amount of data is large enough .
2. More intuitive data , Corresponding SQL Statements are easier to write .
3. More flexible , No matter what new data needs to be stored , All nodes are uniform , Only node attributes and edge attributes need to be considered .
4. The operation of the database will not decrease significantly with the increase of the database .
inferiority ：1. Extremely slow insertion speed .
2. Oversized nodes . When a node has many edges ,
The operation speed of this node will be greatly reduced
9,LR and SVM What are the connections and differences between ?
They are all classification algorithms
If the kernel function is not considered ,LR and SVM Are linear classification algorithms , In other words, their classification decision surfaces are linear .
LR and SVM All supervised learning algorithms
LR and SVM The loss function is different
SVM Only the points near the local boundary line are considered ,LR Consider the overall situation , Distant points also play a role in determining the boundary line .
10, What is clustering , Application scenario of clustering ?
Clustering is based on certain criteria , According to this criterion, a thing is summarized into several parts that do not coincide with each other , In machine learning , Clustering means according to a standard , This criterion is usually similarity , Divide the sample into several parts , Yes, we have to get together with a high degree of similarity , Those with low similarity are separated from each other .
Application scenario of clustering , Perfect job information （ About 10 10000 high-quality resumes , Some resumes contain complete fields , Some resumes are in education , company size , salary , Some fields are empty . Hope to learn from the data , Coding and testing , Dig out the trend and law of position path , Form algorithm model , Predict the empty information in the data .）
The above is about artificial intelligence 10 To interview questions . I will share more about artificial intelligence learning with you later , Interview, etc .
I will continue to share some interview tips Remember to give me some praise !
Technology