Algorithms and Frameworks Used in the
Development of Machine Learning Models
This slide introduces the topic of the presentation, which covers the commonly used machine learning algorithms and the Python frameworks that can be utilized to implement them.
Supervised Learning Algorithms
Linear Regression Random Forests
Used for predicting a continuous value by learning the An ensemble method that uses many decision trees to
relationship between input features and target variable. improve classification or regression performance by reducing
overfitting and increasing stability.
Logistic Regression
Used for binary classification tasks, such as spam detection or Gradient Boosting Machines (GBM)
loan approval, by learning the probability of a binary outcome. Another ensemble technique that builds decision trees in a
sequential manner, correcting the errors of the previous trees
to improve overall performance.
Support Vector Machines (SVM)
Effective in high-dimensional spaces, used for both regression
and classification tasks by finding the optimal hyperplane that Neural Networks
maximizes the margin between classes. Highly flexible models capable of handling various types of
data and tasks, including image and speech recognition, by
learning complex patterns in data.
Decision Trees
Used for both classification and regression, forming a tree-like
model of decisions based on feature values to make K-Nearest Neighbors (KNN)
predictions. A simple algorithm that stores all available cases and classifies
new cases based on a similarity measure, such as distance
functions.
Unsupervised Learning Algorithms
K-Means Clustering Independent Component
Automatically partition a dataset into K distinct Analysis (ICA)
groups based on feature similarities
Similar to PCA but finds components that are
maximally independent of each other
Hierarchical Clustering
Builds a tree of clusters, particularly useful for Autoencoders
hierarchical data like taxonomies
Neural network-based models used for learning
efficient codings of unlabeled data
Principal Component Analysis
(PCA) Generative Adversarial Networks
Technique used to emphasize variation and bring (GANs)
out strong patterns in a dataset
An approach where two models are trained
simultaneously; one generates candidates and the
other evaluates them
Reinforcement Learning Algorithms
Q-Learning Actor-Critic Methods
A model-free reinforcement learning algorithm that Combine the benefits of value iteration and policy
learns the value of an action in a particular state. gradient methods by using two models: one for the
policy and one as the value estimator.
Deep Q Network (DQN)
Combines Q-Learning with deep neural networks to Monte Carlo Tree Search (MCTS)
approximate the Q-value functions. Used in decision processes for calculating statistics
by averaging over sample paths at a decision node.
Policy Gradient Methods
Learn a parameterized policy that can be optimized A3C (Asynchronous Advantage
directly. Actor-Critic)
Uses deep learning and a variant of actor-critic
methods with a parallel architecture to improve
training speed and stability.
Top Python Frameworks
Scikit-learn LightGBM
A comprehensive machine learning library that provides efficient implementations A gradient boosting framework that uses tree-based learning algorithms, designed
of a wide range of supervised and unsupervised learning algorithms, including for efficiency and speed, particularly suitable for handling large-scale data and
linear regression, logistic regression, support vector machines, decision trees, and high-dimensional features.
clustering methods.
CatBoost
TensorFlow A gradient boosting framework that specializes in handling categorical data,
A powerful open-source library for numerical computation and large-scale machine providing state-of-the-art results for standard machine learning tasks.
learning, with a focus on deep learning models and neural networks. It provides a
flexible ecosystem of tools, libraries, and community resources.
OpenAI Gym
A toolkit for developing and comparing reinforcement learning algorithms,
Keras providing a wide variety of environments for training agents and testing various RL
A high-level neural networks API that runs on top of TensorFlow, providing a user- approaches.
friendly interface for building and training deep learning models. It simplifies the
development process and is suitable for both beginners and experienced users.
Stable Baselines
An extension of the original OpenAI baselines library, providing reliable and
PyTorch improved implementations of popular reinforcement learning algorithms, such as
A open-source machine learning library that emphasizes flexibility and research, A3C, PPO, and TRPO.
with a focus on deep learning applications. It is particularly popular in the research
community for its dynamic computation graph and ease of use.
XGBoost
A highly efficient and scalable implementation of the gradient boosting algorithm,
known for its performance in structured data problems, such as classification and
regression tasks.
