All Machine
Learning
Algorithms
EXPLAINED IN
ONE LINE
@RAMCHANDRAPADWAL
Supervised Learning
Algorithms
LINEAR REGRESSION
Predicts a continuous output
variable based on linear
relationships between input
features.
LOGISTIC REGRESSION
Classifies input data into discrete
categories using a logistic
function to model the probabilities.
DECISION TREES
Constructs a tree-like model by
splitting data based on
features to make decisions or
predictions.
RANDOM FORESTS
Ensemble method that combines
multiple decision trees to improve
prediction accuracy and reduce
overfitting.
@RAMCHANDRAPADWAL
SUPPORT VECTOR MACHINES
Separates data into different
classes by finding an optimal
hyperplane in a high-dimensional
space.
NAIVE BAYES
Uses Bayes’ theorem and assumes
independence between features
to classify data based on
probability calculations.
K-NEAREST NEIGHBORS
(K-NN)
Classifies data based on the
majority vote of its k nearest
neighbors in the feature space.
GRADIENT BOOSTING
ALGORITHMS
Ensemble methods that
sequentially build weak models,
minimizing the errors of previous
models to improve predictions.
@RAMCHANDRAPADWAL
Unsupervised Learning
Algorithms
K-MEANS CLUSTERING
Divides data into k clusters based
on similarity, aiming to minimize
the intra-cluster variance.
HIERARCHICAL CLUSTERING
Builds a hierarchy of clusters by
iteratively merging or splitting
them based on similarity.
DBSCAN
Density-based clustering
algorithm that groups together
data points in high-density regions
while marking outliers as noise.
GAUSSIAN MIXTURE MODELS
(GMM)
Models data as a combination of
Gaussian distributions to perform
probabilistic clustering.
@RAMCHANDRAPADWAL
PRINCIPAL COMPONENT
ANALYSIS (PCA)
Reduces the dimensionality of
data by transforming it into a new
set of uncorrelated variables
called principal components.
T-DISTRIBUTED STOCHASTIC
NEIGHBOR EMBEDDING (T-SNE)
Dimensionality reduction
technique that visualizes high-
dimensional data in a lower-
dimensional space, emphasizing
local structure.
@RAMCHANDRAPADWAL
Semi - Supervised
Learning Algorithms
EXPECTATION-MAXIMIZATION
(EM)
Iteratively estimates the
parameters of a probabilistic model
by alternately computing
expected values and maximizing
likelihood.
SELF-TRAINING
Uses a small amount of labeled
data to train a model, which is then
used to label a larger amount of
unlabeled data for further training
iterations.
CO-TRAINING
Simultaneously trains multiple
models on different subsets of
features or data instances,
leveraging their agreement on the
unlabeled data.
LABEL PROPAGATION
Propagates labels from labeled
instances to unlabeled instances
based on their similarity, utilizing
the local structure of the data.
GENERATIVE MODELS WITH
LABELED AND UNLABELED DATA
Combines generative models with
both labeled and unlabeled data to
estimate class distributions and
make predictions.
@RAMCHANDRAPADWAL
Reinforcement
Learning Algorithms
Q-LEARNING
Reinforcement learning algorithm
that learns through trial and error,
optimizing actions based on
maximizing cumulative rewards.
DEEP Q-NETWORK (DQN)
Reinforcement learning algorithm
that combines Q-learning with
deep neural networks for improved
performance in complex
environments.
PROXIMAL POLICY
OPTIMIZATION
Policy Optimization algorithm that
iteratively updates policies to
maximize rewards and improve
sample efficiency.
MONTE CARLO TREE SEARCH
(MCTS)
Search algorithm that simulates
and evaluates possible moves in a
game tree to determine optimal
actions
ACTOR-CRITIC METHODS:
Reinforcement learning approach
that combines a policy network
(actor) and a value function
(critic) to guide learning.
@RAMCHANDRAPADWAL
Deep Learning
Algorithms
CONVOLUTIONAL NEURAL
NETWORKS (CNN)
Deep learning models designed for
image processing, using
convolutional layers to extract
meaningful features.
RECURRENT NEURAL NETWORKS
(RNN)
Neural networks that can
process sequential data by
retaining and using information
from previous inputs.
LONG SHORT-TERM MEMORY
(LSTM)
A type of RNN that addresses the
vanishing gradient problem and
can retain information over longer
sequences.
GENERATIVE ADVERSARIAL
NETWORKS (GAN)
Neural network architecture
consisting of a generator and a
discriminator, trained in
competition to produce realistic
data.
@RAMCHANDRAPADWAL
TRANSFORMER NETWORKS
Architecture that employs self-
attention mechanisms to process
sequences, widely used in natural
language processing tasks.
AUTOENCODERS
Neural networks designed to
learn compressed
representations of input data
by training to reconstruct the
original input from a reduced-
dimensional representation.
@RAMCHANDRAPADWAL
Ensemble Learning
Algorithms
BAGGING
Ensemble technique that
combines multiple models trained
on different subsets of the
training data to make predictions.
BOOSTING
Ensemble method that combines
weak learners sequentially, with
each subsequent model focusing
on instances that previous
models struggled with.
STACKING
Ensemble approach that combines
predictions from multiple models
by training a meta-model on their
outputs.
VOTING CLASSIFIERS
Ensemble method that combines
predictions from multiple models
by majority voting or weighted
voting.
@RAMCHANDRAPADWAL
Dimensionality
Reduction Algorithms
PRINCIPAL COMPONENT
ANALYSIS (PCA)
Reduces the dimensionality of
data by transforming it into a new
set of uncorrelated variables
called principal components.
LINEAR DISCRIMINANT ANALYSIS
(LDA):
Maximizes class separability by
finding linear combinations of
features that best discriminate
between classes.
INDEPENDENT COMPONENT
ANALYSIS (ICA)
Separates a multivariate signal
into additive subcomponents to
discover underlying independent
sources.
VARIATIONAL AUTOENCODERS
(VAE)
Neural network-based generative
models that learn low-dimensional
representations and reconstruct
original data with high fidelity.
@RAMCHANDRAPADWAL
Transfer
Learning Algorithms
PRE-TRAINED DEEP NEURAL
NETWORKS
Deep learning models that are
trained on large-scale datasets for
specific tasks, often used as a
starting point for transfer learning.
FINE-TUNING
Technique where a pre-trained
model is further trained on a
specific task or dataset to
improve its performance.
DOMAIN ADAPTATION
Technique that transfers
knowledge from a source domain
to a target domain with different
distributions, improving
generalization.
MULTI-TASK LEARNING
Simultaneously trains a model
on multiple related tasks to
improve overall performance by
leveraging shared information.
@RAMCHANDRAPADWAL
