Disruptive Technologies

LECTURE 2

Artificial Intelligence and Machine Learning

Dr. Tamal Ghosh, Department of Computer Science & Enginee

Definition of Machine Learning (ML)
Machine Learning (ML) is a subfield of artificial intelligence
(AI) that focuses on the development of algorithms and
models that enable computers to learn and make predictions
or decisions without being explicitly programmed.
ML systems use data to improve their performance over
time.
 Distinction Between Ai And Ml
AI ML

Definition: AI, or Artificial Intelligence, is a broad Definition: ML, or Machine Learning, is a subset
concept referring to the development of machines of AI that focuses specifically on the development
or systems that can perform tasks that typically of algorithms and models that enable computers
require human intelligence. These tasks include to learn and make predictions or decisions based
learning, reasoning, problem-solving, on data.
understanding natural language, and perception. Scope: ML is a more specialized field within AI,
Scope: AI encompasses a wide range of emphasizing the use of statistical techniques to
capabilities, from basic rule-based systems to allow computers to improve their performance on
highly advanced systems capable of complex a specific task without being explicitly
decision-making and learning from data. programmed.
Applications: AI applications include speech Applications: ML applications include
recognition, image processing, natural language recommendation systems, image and speech
understanding, game playing, and autonomous recognition, predictive analytics, and personalized
systems such as self-driving cars. marketing.
Goal: The ultimate goal of AI is to create Goal: The primary goal of ML is to enable
machines that can perform any intellectual task a machines to learn patterns from data and make
human can, exhibiting general intelligence. predictions or decisions without being explicitly
programmed for each task.
 Key differences
Scope: AI is a broader concept encompassing various approaches to simulate
human intelligence, while ML is a specific subset of AI focused on learning from
data.
Learning: AI systems may or may not involve learning. Some AI systems rely
on rule-based logic, while others incorporate learning capabilities. ML, on the
other hand, specifically emphasizes learning from data.
Programming: In traditional AI, systems are explicitly programmed with rules
and logic. In ML, the emphasis is on developing algorithms that can learn
patterns and make decisions based on data, reducing the need for explicit
programming.
Flexibility: AI systems can be rule-based or learning-based, providing a
broader range of approaches. ML systems, by definition, are learning-based
and rely on data to improve their performance.
Examples: AI includes a wide range of applications, from rule-based systems
like expert systems to complex learning systems. ML examples include
 Types of Machine Learning
Supervised Learning:
In supervised learning, the algorithm is trained on a labeled dataset, where the
input data is paired with corresponding output labels. The goal is to learn a
mapping from inputs to outputs so that the algorithm can make predictions on
new, unseen data.
Unsupervised Learning:
Unsupervised learning involves training the algorithm on an unlabeled dataset.
The algorithm must discover patterns, relationships, or structures within the
data without explicit guidance on the output. Clustering and dimensionality
reduction are common tasks in unsupervised learning.
Reinforcement Learning:
Reinforcement learning involves an agent learning to make decisions by
interacting with an environment. The agent receives feedback in the form of
rewards or punishments based on its actions, allowing it to learn optimal
 The Role of Data in Machine Learning
Data plays a central and critical role in the field of Machine Learning
(ML). The entire process of training, evaluating, and deploying
machine learning models is fundamentally dependent on the quality,
quantity, and relevance of the data used. Key aspects of role of data
in ML:

1. Training Data:
Definition: Training data is the set of examples used to train a
machine learning model. It consists of input-output pairs where the
model learns the mapping between inputs and corresponding outputs.
Importance: The quality and representativeness of the training data
directly impact the performance and generalization ability of the
model. Diverse and well-labeled data contribute to better learning and
 Contd…
2. Feature Selection and Engineering:
Definition: Features are the variables or attributes in the data that influence
the model's predictions. Feature selection and engineering involve choosing
relevant features or creating new ones to improve model performance.
Importance: The right features enhance the model's ability to capture
patterns and make accurate predictions. Effective feature engineering can
significantly impact the model's efficiency.
3. Data Preprocessing
Definition: Data preprocessing involves cleaning and transforming raw data
into a format suitable for training. This step may include handling missing
values, normalizing data, and addressing outliers.
Importance: Proper preprocessing ensures that the data is in a suitable form
for the machine learning algorithm, preventing issues such as biased models
or inaccuracies due to data inconsistencies.
 Contd…
4. Model Evaluation and Testing:
Definition: After training, models need to be evaluated using a separate set
of data not seen during training (test set). This evaluation measures the
model's performance and generalization to new, unseen data.
Importance: Testing data is crucial for assessing the model's ability to make
accurate predictions on new instances. It helps identify overfitting and
ensures the model's reliability.

5. Unbiased and Representative Data:

Definition: Unbiased and representative data reflects the true distribution of
the real-world scenarios the model will encounter. Biased data can lead to
discriminatory or inaccurate models.
Importance: To ensure fair and unbiased predictions, it's essential to use
data that accurately represents the diverse range of conditions the model will
Contd…
6. Continuous Learning and Adaptation:
Definition: Continuous learning involves updating a model over time as new data
becomes available. This adaptability is crucial for ensuring models remain relevant
and accurate in evolving environments.
Importance: Regular updates with fresh data help models adapt to changes, such as
shifts in user behavior, preferences, or other dynamic factors.

7. Ethical Considerations:
Definition: Ethical considerations involve ensuring that the data used for training
and testing is collected and utilized responsibly, without reinforcing biases or
causing harm.
Importance: Ethical data practices are vital for developing fair and trustworthy
machine learning models. Awareness of potential biases and privacy concerns is
essential to building responsible AI systems.
Evaluation Metrics In Ml
Classification Metrics:
 Accuracy: Definition: The proportion of correctly classified instances out of the
total instances.
Use Case: Suitable for balanced datasets.
 Precision: Definition: The ratio of true positive predictions to the total
predicted positives.
Use Case: Relevant when minimizing false positives is crucial.
 Recall (Sensitivity or True Positive Rate): Definition: The ratio of true
positive predictions to the total actual positives.
Use Case: Important when minimizing false negatives is a priority.
 F1 Score: Definition: The harmonic mean of precision and recall.
Use Case: Balances precision and recall in situations where there is an
imbalance between classes.
Contd…
Regression Metrics:
Mean Absolute Error (MAE): Definition: The average absolute
difference between predicted and actual values.
Use Case: Provides a simple and interpretable measure of prediction
error.
Mean Squared Error (MSE): Definition: The average squared
difference between predicted and actual values.
Use Case: Sensitive to outliers; commonly used but not robust to extreme
values.
Root Mean Squared Error (RMSE): Definition: The square root of the
average squared difference between predicted and actual values.
Use Case: Similar to MSE but in the original units of the target variable.
Contd…
Clustering Metrics:

Silhouette Score: Definition: Measures how similar an object is to its own

cluster compared to other clusters.
Use Case: Evaluate the quality of clustering.
Davies-Bouldin Index: Definition: Measures the compactness and separation
between clusters.
Use Case: Lower values indicate better clustering.

Choosing the appropriate evaluation metric is crucial for understanding how well
a machine learning model is performing in a given context. It ensures that the
model is aligned with the specific goals and requirements of the problem at hand.