Fundamentals of Mathematics, Engineering ability, feature Engineering, model evaluation, optimization algorithm, basic concepts of machine learning, classical machine learning model, deep learning model, business and applications
directory
1. Mathematical foundation
1.1. Probability Theory
1.2. Linear Algebra
1.3. Calculus
1.4 convex optimization
1.5. Information theory
Ii. Engineering ability
2.1. Data structure and algorithm
(1) Tree and correlation algorithm
(2) Graphs and related algorithms
(3) Hash table
(4) Matrix operation and optimization
2.2 big data processing
(1), and graphs
(2), the Spark
(3), HiveQL
(4), Storm
2.3 machine learning platform
(1), TensorFlow
(2), the Torch
(3), Theano
2.4. Parallel computing
2.5. Databases and data warehouses
2.6 System service architecture
3. Characteristic engineering
3.1. Feature discretization and normalization
3.2. Feature combination
3.3. Feature selection
3.4 word embedding representation
4. Model evaluation
4.1 Evaluation indicators
4.2 A/B test
4.3 Overfitting and underfitting
4.4. Super parameter selection
5. Optimization algorithm
5.1. Loss function
5.2. Regularization
5.3. EM Algorithm
5.4 Gradient descent/Stochastic gradient descent
5.5. Back propagation
5.6 gradient verification
5.7, Momentum
5.8, AdaGrad
5.9, Adam
Basic concepts and classification of machine learning
6.1 Basic Concepts
(1) Hypothesis space
(2) Training/test data
(3), mark
(4) Loss function
6.2. Classification by data
(1), classification
(2), regression
(3) Sequence annotation
6.3 Classification according to supervision
(1) Supervised learning
(2) Unsupervised learning
(3) Reinforcement learning
6.4. Classification by model
(1) Generative model
(2) Discriminant model
Classic machine learning model
7.1 Supervised learning
(1) Classical algorithm
(2) Probability graph model
7.2. Unsupervised Learning
(1) Hierarchical clustering
(2) K-means clustering
(3) Gaussian mixture model
(4) Theme model
7.3. Integrated learning
(1), Bagging
(2), Boosting
(3), GBDT
(4) Random forest
7.4. Dimension reduction algorithm
7.5, sample
7.6. Reinforcement learning
Deep learning model
8.1. Forward neural network
(1) Multi-layer perceptron
(2) Convolutional neural network
(3) Deep residual network
(4) Self-organizing mapping neural network
(5) Limited Boltzmann machine
8.2. Recurrent neural network
(1) Cyclic neural network
(2) Long and short-term memory model
(3) Attention mechanism
(4), Seq2Seq
8.3 optimization skills of deep learning
(1) Batch normalization
(2), Dropout
(3) Activation function
8.4. Reinforcement learning
8.5 Generative adversarial network
Ix. Business and Application
9.1. Computer Vision
9.2. Natural language processing
9.3. Recommendation System
9.4. Calculate advertising
9.5. Smart games
Related article AI: A few diagrams clarify the ambiguous relationship between ARTIFICIAL intelligence and machine learning, knowledge discovery, data mining, statistics, pattern recognition, neurocomputing, and databases
1. Mathematical foundation
Machine Learning and Advanced Mathematics: Basic Concepts, Code Implementation, and Case Applications — Advanced DL simpleNet Use custom simpleNet(set weights) to predict, evaluate, and output gradients for new samples
1.1. Probability Theory
- Common probability distribution
- The large number theorem and the central limit theorem
- Hypothesis testing
- Bayes theory
1.2. Linear Algebra
1.3. Calculus
1.4 convex optimization
1.5. Information theory
Ii. Engineering ability
2.1. Data structure and algorithm
【Algorithm 】 the Algorithm of the Algorithm of the path of progress 】 ten classical sorting Algorithm Algorithm: Algorithm: Data structure related exercises (array, string, linked list, stack, queue, tree, graph, hash) Mathematical programming (time speed, base conversion, permutation and combination, conditional probability, Fibonacci sequence) in the Algorithm of mathematical programming (time speed, base conversion, permutation and combination, conditional probability, Fibonacci sequence) The machine learning Algorithm is based on the Algorithm of searching, sorting, recursion, complexity, and advanced Algorithm. The Algorithm is an advanced approach to Python
(1) Tree and correlation algorithm
(2) Graphs and related algorithms
(3) Hash table
(4) Matrix operation and optimization
2.2 big data processing
BigData: introduction to BigData development, core knowledge (Linux basics +Java/Python programming language +Hadoop{HDFS, HBase, Hive}+Docker), detailed introduction to classic scenarios Artificial intelligence & Big data related positions — introduction, skills, benefits and advanced detailed introduction of [data analyst]
(1), and graphs
(2), the Spark
(3), HiveQL
(4), Storm
2.3 machine learning platform
Relevant article DL framework: the mainstream deep learning framework (TensorFlow/Pytorch/Caffe/Keras/CNTK/MXNet/Theano/PaddlePaddle) introduction, comparison and case application of multiple orientations detailed strategy MXNet of DL framework: Deep learning framework MXNet introduction, installation, use method, application case details DL framework Caffe: DL: Pytorch, Tensorflow DL: Tensorflow DL: Pytorch, Tensorflow DL: Tensorflow Tensorflow is a framework for deep learning. Tensorflow Core(Tensorflow Core API) is a framework for deep learning. Tensorflow is a framework for deep learning. Darknet: A deep learning framework called PyTorch is an AutoKeras framework that allows you to install and use PyTorch. Keras is a framework for deep learning. It is a framework for deep learning. Keras is a framework for deep learning
(1), TensorFlow
(2), the Torch
(3), Theano
2.4. Parallel computing
2.5. Databases and data warehouses
SQLSever: SQLSever Database management study and in-depth understanding of SQL command statement advanced comprehensive chapter “primary → intermediate → advanced” (continue to update, recommend collection)
2.6 System service architecture
3. Characteristic engineering
Abstract: the DataScience task (data analysis, feature engineering, scientific prediction, etc.) based on machine learning is one of the most important DataScience tasks in the world. ML FE: Data Processing-An introduction to feature engineering, a detailed introduction to feature engineering, a detailed introduction to use, and a detailed overview of case applications
3.1. Feature discretization and normalization
3.2. Feature combination
3.3. Feature selection
3.4 word embedding representation
4. Model evaluation
Model evaluation index (loss function) for ML: collection of Scoring/metrics based on algorithms in different Machine learning frameworks (SKlearn /TF) (code only)
4.1 Evaluation indicators
4.2 A/B test
4.3 Overfitting and underfitting
DNN optimization techniques for DL: introduction, usage and case study of DNN techniques to suppress over-fitting/under-fitting and improve generalization ability
4.4. Super parameter selection
DL Model tuning: Optimization parameters of deep learning algorithm Model Tuning using grid Search for hyperparameters of deep learning model
5. Optimization algorithm
5.1. Loss function
Lf-ml: Gradient/derivation of loss functions (LiR loss, L1 loss, L2 Loss, Logistic Loss) in Machine learning A brief introduction of the common loss function (continuous/discrete) in machine learning, the difference between loss function/cost function/objective function, and a detailed overview of case applications
5.2. Regularization
AI: Neural network parameters (data, layer number, batch size, learning rate + activation function + regularization + classification/regression) and result visualization
5.3. EM Algorithm
5.4 Gradient descent/Stochastic gradient descent
5.5. Back propagation
5.6 gradient verification
5.7, Momentum
DL DNN optimization technology: GD, SGD, Momentum, NAG, Ada series, RMSProp various code implementation detailed walkthrough
5.8, AdaGrad
5.9, Adam
Basic concepts and classification of machine learning
6.1 Basic Concepts
(1) Hypothesis space
(2) Training/test data
(3), mark
(4) Loss function
6.2. Classification by data
(1), classification
(2), regression
(3) Sequence annotation
6.3 Classification according to supervision
SL: Supervised Learning: The concept, application, and SSL of Unsupervised Learning Semi-Supervised Learning: a brief introduction, application and a detailed introduction of classic cases
(1) Supervised learning
(2) Unsupervised learning
(3) Reinforcement learning
6.4. Classification by model
(1) Generative model
(2) Discriminant model
Classic machine learning model
7.1 Supervised learning
(1) Classical algorithm
Support vector machine
Logistic regression
The decision tree
(2) Probability graph model
Naive Bayes
Maximum entropy model
Hidden Markov model
Conditional random field
7.2. Unsupervised Learning
The Clustering algorithm of ML: the introduction of the Clustering algorithm, the main ideas, key steps, code implementation and other relevant detailed introduction
(1) Hierarchical clustering
(2) K-means clustering
(3) Gaussian mixture model
(4) Theme model
7.3. Integrated learning
EL: Ensemble Learning (Ensemble Learning) concept introduction, problem application, algorithm classification, key steps, code implementation, etc
(1), Bagging
(2), Boosting
(3), GBDT
(4) Random forest
7.4. Dimension reduction algorithm
Linear dimensionality reduction of FE and DR: PCA/ whitening, mathematical knowledge of LDA algorithm (covariance matrix), related papers, algorithm, code implementation, case application and other related illustrations
7.5, sample
DataScience: Introduction and summary of experience for severely imbalanced data set using various sampling strategies (stochastic over-sampling, SMOTE over-sampling, SMOTETomek comprehensive sampling, change of sample weight, etc.)
7.6. Reinforcement learning
ML RL: Reinforcement Learning introduction, application, classic cases, and a detailed guide to Learning resources
Deep learning model
8.1. Forward neural network
(1) Multi-layer perceptron
DL Perceptron: Introduction, principle and case application of Perceptron (Perceptron/multilayer Perceptron/artificial neuron)
(2) Convolutional neural network
DL CNN: Introduction to convolutional neural network algorithms for computer vision (classical architecture/paper), CNN optimization techniques, paramedic learning practices, CNN classical structure and evolution, a detailed walkthrough of case applications
(3) Deep residual network
DL ResNet: a brief introduction of ResNet algorithm (paper introduction), a detailed explanation of the architecture, case application, etc
(4) Self-organizing mapping neural network
(5) Limited Boltzmann machine
8.2. Recurrent neural network
(1) Cyclic neural network
DL RNN: An introduction to RNN for recurrent neural networks, applications and a detailed walkthrough of classic cases
(2) Long and short-term memory model
LSTM of DL: Introduction to LSTM algorithm (principle, key steps, RNN/LSTM/GRU comparison, single and multi-layer LSTM), a detailed walkthrough of case application
(3) Attention mechanism
DL Attention: An introduction to the Attention mechanism of Attention and a detailed overview of its application fields
(4), Seq2Seq
8.3 optimization skills of deep learning
(1) Batch normalization
DataScience: To explore and analyze the difference and relationship between standardized standardization and normalized Normalization/ scaled Scaling of data processing in machine learning
(2), Dropout
DNN optimization techniques for DL: Use Dropout optimization methods to improve the performance of DNN models
(3) Activation function
- Sigmoid
- Softmax
- Tanh
- ReLU
ML/DL activation function/derivative function: COMMONLY used IN ML AF activation function (step_function, sigmoid, softmax, ReLU, etc.)& derivative function code implementation detailed walkthrough
8.4. Reinforcement learning
ML RL: Reinforcement Learning introduction, application, classic cases, and a detailed guide to Learning resources
8.5 Generative adversarial network
DL GAN: generates an introduction, application, and detailed walkthrough of classic cases against network GAN
Ix. Business and Application
9.1. Computer Vision
CV: An introduction to computer vision in artificial intelligence (CV history + common data sets +CV positions), a comparison of traditional methods with CNN algorithms, a detailed overview of eight applications of computer vision (knowledge mapping + classic cases)
9.2. Natural language processing
NLP: introduction to natural language processing, history, and case studies
9.3. Recommendation System
RS of ML: Recommendation system based on CF+LFM (film recommendation based on highly relevant users)
9.4. Calculate advertising
FE in ML: Data Processing — A Case study of Feature Engineering in High dimensional Combination feature Processing (Matrix Decomposition) — Advertising click estimation problem based on LoR Algorithm
9.5. Smart games
RL PG: Based on TF using strategy gradient algorithm to play Cartpole game to achieve intelligence