AI is a complex subject and hard to learn
Often, in the early stages, people make mistake such as
a) They try to learn everything
b) They do not know in which order to learn
c) They go to deep into one subtopic initially
Hence, I created a minimum viable learning framework for self learning AI (machine learning and deep learning)
Because its concise and its minimal, it does not include topics like GANs. Reinforcement learning etc. tt also does not cover Bayesian approaches n detail
However, this syllabus should get you to about 80 percent of your journey for a typical data science role
Statistics
Central limit theorem
Sampling methods
Type i vs type ii error
Selection bias
Non gaussian distributions
Bias variance tradeoff
Confusion matrix
Normal distribution
Correlation
Covariance
Point estimates and confidence interval
a/b testing
p-value
re-sampling
Methods to overcome combat overfitting and underfitting
Treatment of outliers
Treatment of missing values
Confounding variables
Entropy and information gain
Cross validation
Basic concepts
Between a validation set and a test set
Supervised learning
Unsupervised learning
Parameters v.s. hyperparameters
Cost function
Regression
Linear regression
Assumptions required for linear regression
Limitations of linear regression
Deep learning
What is the difference between machine learning and deep learning?
Basic working of Neural networks
Soft-max
Relu
Learning rate
Epoch / batch and iteration
The convolution operation
Layers of a CNN
Pooling operation
Kernels and Parameter sharing
Back propagation
Gradient descent
Vanishing gradients
Activation functions
LSTM
Models
Regression
Classification
logistic regression
SVM
Tree based
Clustering
PCA / Dimensionality reduction
MLP
CNN
Autoencoders
Regularization
Lasso
Ridge
Regularization in deep learning (ex dropout)
Ensemble methods
Boosting
Bagging
Optimization techniques
Matrix optimization techniques (contrast to)
Gradient descent including specific optimizers like Adam etc
Back propagation
Statistical inference
Models
Parametric models
Non-Parametric models
Paradigms
Frequentist
Bayesian
Statistical proposition/outcome
A point estimate
An interval estimate
A credible interval
rejection of a hypothesis
Clustering or classification of data points into groups.
Parameter Estimation techniques
Ordinary least square estimation
Maximum likelihood estimators
Hyperparameter tuning techniques
Grid search
Random search
Feature Engineering
Feature Extraction (ex PCA)
Feature Transformation (ex binning, log transforms)
Feature Selection (filter methods, wrapper methods etc)
Model evaluation
Regression metrics
(R)MSE
MAE
R²
Classification metrics
Accuracy
Recall
Precision
F1 score
Confusion matrix
Hope you find it useful
){kind=link}