AI Top 30 Course Mind Map

AI Top30
Programming Languages
Programming
R
Juila
Hello World
Flow Control
Functions
Data Types
File Handling
Objects and Classes
Date and Time
Python
Libraries
Numpy
Pandas
Series
DataFrame
Array
Math for Machine Learning
Linear Algebra
Calculus
Probability and Statistics
Matrices
Determinants
Eigen Values and Vectors
Derivatives
Chain Rule
Distributions
Visualization
Machine Learning
Matplotlib
Seaborn
Plotly
Minima and Maxima
Gradient Descent
Linear Regression
Logistic Regression
Deep Learning
Fundamantals
Bias and Variance
Overfitting and Underfitting
Cross Validation
Test Train Split
Types
Supervised
Unsupervised
Reinforcement
Libraries
Types
Line
Area
Histogram
Bar
Pie
Box
Scatter
Bubble
Folium
Wordclouds
Transpose
Symmetric Matrix
Inverse
Solving linear equation
Variance
Covariance
Uivariate
Multivariate
Basic ML Steps
Polynomial
Binary Classification
Multi-class Classification
Import Libraries
Import Data
Data Cleaning
Correlation and find dominant features
Test Train Split
Model Training
Evaluation Results
Libraries
Data Manipulation
Algorithms
Map
Feature Engineering
Tree Based Algorithms
Decision Trees
Random Forest
Types
ID3
C4.5
CART
Building Tree
Gini Impurity
Gini Gain
Entrophy
Information Gain
Bagging
Boosting
Bootstraped Dataset
Aggregate the Results
Decision Tree Regressor
Data Preprocessing
Root, Internal, Leaf Nodes
Decision Node
Yes/No
Numeric (Sort and Average of Consecutive Values)
Ranked Data
Multiple Choice Data
Calc Impurity for each one and combination
Datasets
Out of Bag Dataset/Error
Randomly choose variables for decision nodes impurity
AdaBoost
Gradient Boost
XGBoost
Ensemble MEthods
Weighted Dataset
Learns from previous mistake
sklearn
public
Iris
Police violation
Medical Appointment
Kaggle
Car Dataset
Old Car Price Dataset
Titanic
Salary Prediction
House Price Prediction
Encoding
One-Hot
Integer Encoding
Label Encoding
Handling Mission Values
Remove rows with null values
Impute with mean or median
Test and Train Split
Feature Scaling
Normalization
Standardization
Binning (continuous to categorical)
Dimensionality Reduction
SVM
K-Means
DBSCAN
Hierarchical Clustering
Dimensionality REduction
PCA
LDA
Intro
Tensorflow
Feed Forward NN/Shallow Networks
Deep NEural Networks
CNN
AlexNet
VGGNet
ResNet
Inception NEt
Residual Blocks/Skip connections
Inception Block
Transfer Learning
Datasets
Fashion MNIST
Digit MNIST
Imagenet
Custom Images
Entire Model
Feature Extration/Retrain
NLP
Bag of words
Tokenizer
Word2Vec
Data Cleaning
Stemming
lem
stop word removal
n-grams
Sequence Models
RNN
LSTM
GRU
Vanishing GRadient
Encoder Decoder Model
Input -> One word at a time
Cannot use GPU/Parrallelize
Language TRanslation
Attention
Transformers
Positional Encoding
Multi Headed Self Attention
Easily PArallize
Input -> Entire Sentence
Hugging Face
Adv Computer Vision
Object Localization
1 class
Object Detection
Multi Class
Types
RCNN
Fast RCNN
Faster RCNN
SSD
YOLO
Segmentation
Instance
Semantic
UNET
Mask RCNN
Deployment
Pickle/Flask
MLflow
TF Serve
Null Value
Remove
Impute
Mean, Median
Frequency, Mode
Model -> Function Mapping
X1 -> X2
Data Type Setting
Outliers Removal
Normalization
Continuous
Categorical
Standardization
0 to 1
Min Max Scalar
Divide by Max Number
(x - min)/(max - min)
x/max
mean centering
Binning
Continuous to Categorical
Label Encoding
One-Hot Encoding
Based on Data Type
Date
String
Day of a week - Weekend
Polynomial Feature Transformation
PCA
Feature Selection
ANOVA
Covariance
Range - infy to + infy
Range -1 to 1
AUC
ROC
F1-score (aka F-Score / F-Measure)
Precision
Accuracy
Specificity
Sensitivity/Recall
TP, TN, FP, FN
Confusion Matrix
Classification
R-Squared Score
RMSE
MAE
MSE
Regression
Evaluation Metrics
Hyperparameter Tuneing
Cross-Validation
GridSearch
Parallel Tree Building Peocess
Sequential Tree Building PRocess
Decision Stumps
Weak Learners
1. Root Node, 2. Calc Residuals, 3. Build Tree using Residuals, 4. Model + (LR * Tree_Op)
Same as Gradient Boost
Unique Tree Building Processs
Cache Optimization
Parallelisation
DMatrix
Max-Margin Classifier
Choose Edge Vectors
Soft-Margin Classifier
Kernel Trick
Linear Kernel
Polynomial Kernel
Sigmoid Kernel
RBF (RAdial BAsis Function)
Tunable Params { C, gamma }
Resilent to Noise
Dendrogram
Naive Bayes
Bayes Theorem
Conditional PRobability
K - Nearest Neighbours
Why?
Equal Importance to all features
Model Converges Quickly (Gradient Descent)
Integer Encoding
Month - Festival
Recepie - Chicken, Tomato, Cream
Is_chicken
x1, x2 -> x1^2, x2^2, X1*X2, x1, x2
Dimensionality reduction
Memory save
Avoids Underfitting/Underfitting
Visualization
Model Selection
Spot Checking ML Algorithms
Compare models using cross validation
Sub Topic
age (10 to 90) -> Y, A, E, O
Data Distribution
Computational complexity reduced
Cancer
279 2