SESSION - JULY 2024
SEMESTER – 3rd
COURSE – MASTER OF COMPUTER APPLICATIONS (MCA)
SUBJECT – Soft Computing
SUBJECT CODE – OMCA-235
LIVE SESSION - 2
Recap of the Previous Lecture
• Introduction to Soft Computing
• Hard Computing vs. Soft Computing
• Evolution of Computing
• Soft Computing Constituents
• Conventional to Computational Intelligence
Machine Learning Basics
A field of study that enables computers to learn and improve from experience without being explicitly programmed
automatically.
ML algorithms train computers to make predictions, recommendations, or estimations based on historical data.
Key features of Machine Learning
Learning from Experience: Just like humans, computers learn from past data (experience) to perform tasks better over time.
Data-Driven: ML relies heavily on data collection and patterns to make decisions.
Applications: Used for predictions (e.g., weather forecasting), recommendations (e.g., Netflix or Amazon suggestions), and
more.
Why Machine Learning?
Automates complex decision-making.
Increases efficiency by learning from vast datasets.
Improves outcomes by predicting trends or behaviors .
Building Blocks of Machine Learning
Task
Definition: The main problem or objective that the machine learning model aims to solve.
Examples: Predictions, recommendations, estimations.
Goal: Solve real-world problems like predicting customer preferences or recommending movies.
Experience
Definition: Learning from historical or past data to make future predictions or solve tasks.
Importance: The more data available, the better the model can learn from patterns and improve
accuracy.
Example: A machine learning model learning from previous customer purchases to recommend
products.
Performance
Definition: The ability of the machine to solve the problem and provide the best possible outcome.
Dependence: Performance is directly related to the type of problem and the quality of the data.
Goal: Improve model accuracy and provide better outcomes for the task at hand.
Techniques in Machine Learning
1. Supervised Learning
Description: Labeled data (input and output) is used to train models and predict future events.
Example: Identifying dog images after training the model with labeled dog pictures.
2. Unsupervised Learning
Description: Works with unlabeled data to discover hidden patterns or groupings.
Example: Organizing documents into categories without predefined labels.
3. Reinforcement Learning
Description: Agents learn by interacting with their environment and receiving feedback
(rewards/punishments).
Example: A robot learning to navigate a maze by maximizing positive rewards for correct actions.
4. Semi-supervised Learning
Description: Combines both labeled and unlabeled data to improve accuracy while reducing labeling
costs.
Example: Text document classification with limited labeled data and abundant unlabeled data.
Applications of Machine Learning
Difference between Machine learning and Artificial Intelligence
Applications of AI and Machine Learning
Siri (AI): Siri uses Natural Language Processing (NLP), an AI technique, to understand and respond to
human language. It also uses Machine Learning to improve its ability to understand accents, preferences,
and common queries.
Customer Support Chatbots (AI): Chatbots rely on AI to simulate conversations with users. ML is used to
learn from past interactions, improving the chatbot's ability to answer questions accurately over time.
Expert Systems (AI): These systems simulate human expertise to solve specific problems. Machine
Learning models can help refine expert systems by learning from previous cases and improving their
decision-making.
Online Recommender Systems (ML): While ML is the core technology behind personalized
recommendations (such as Netflix or Amazon), these recommendations contribute to the AI goal of
creating intelligent systems that improve user experience.
Google Search Algorithms (ML): Machine Learning is used to enhance search results based on user
behavior and patterns. This contributes to AI by making the search engine "intelligent" in understanding
what users need.
Facebook Auto Friend Tagging (ML): ML uses facial recognition models to identify friends in photos. The
overarching AI goal is to make the social platform smarter in predicting who is in the photo.
Important questions till now
1. What is the fundamental difference between conventional and computational
intelligence?
2. How does machine learning differ from traditional programming?
3. What are the main types of machine learning algorithms?
Key Concepts in Soft Computing
Soft Computing: A branch of computer science that handles approximate reasoning,
uncertainty, and imprecision to create intelligent, adaptive systems.
Hard Computing: The traditional approach focuses on precise logic and deterministic
algorithms, contrasting with soft computing's flexible approach.
Fuzzy Logic: A mathematical method for managing uncertainty using fuzzy sets and rules,
allowing systems to handle imprecision.
Neural Networks: Brain-inspired computing systems that learn from data to make
predictions or decisions.
Evolutionary Computation: Techniques like genetic algorithms that mimic biological
evolution to optimize problem-solving.
Key Concepts in Machine Learning
Supervised Learning: Learning from labeled data to predict or make decisions.
Unsupervised Learning: Identifying patterns in unlabeled data.
Reinforcement Learning: Learning through interaction with the environment and
feedback from actions.
Data-Driven Approach
A method based on analyzing large datasets to uncover insights and patterns.
Example: Netflix recommendation system, which suggests movies based on
analyzing large volumes of viewer data.
