AI Start-up

ESAB SSC
TEAM
 Examples:

Definition: Artificial
Voice Assistants: Alexa,
Intelligence (AI) is the Siri, Google Assistant can
simulation of human understand and respond to
voice commands.

What is Artificial intelligence processes by

machines, especially
Intelligence? computer systems. These
processes include
Recommendation Systems:
Netflix recommends shows
based on viewing history;
learning, reasoning, and Amazon suggests products
based on past purchases.
self-correction.

Autonomous Vehicles:
Self-driving cars by
companies like Tesla use AI
to navigate and make
decisions.
 Definition: Machine Learning (ML) is a subset of AI
that uses statistical techniques to give computers the
ability to "learn" from data without being explicitly
programmed.

Basic ML Concepts:

• Training Data: The dataset used to teach the model.

What is Machine • Models: Algorithms that process the data to make

predictions or decisions.
Learning? • Predictions: The output generated by the model when
applied to new data.

Example: A spam filter that learns from a labeled

dataset of emails (spam and non-spam) and predicts
whether new emails are spam.
 Impact on Industries: AI and ML
revolutionize industries by enhancing
efficiency, accuracy, and innovation.

Use Cases:
Why AI and ML? • Predictive Maintenance: In manufacturing, AI
predicts equipment failures, reducing downtime. E.g.,
GE uses AI to predict when jet engines need
maintenance.
• Quality Control: AI identifies defects in products,
ensuring high quality. E.g., Using computer vision to
detect imperfections in assembly lines.
• Healthcare: AI assists in diagnosing diseases and
personalizing treatment plans. E.g., IBM Watson for
Oncology helps in cancer diagnosis.
 • Chatbots: AI-powered chatbots handle
customer queries efficiently. E.g., Chatbot
Custo on websites providing 24/7 customer
mer
Service
support.
:

Applications of AI • Recommendation Systems: Platforms like

and ML Person
alizatio
Netflix and Amazon personalize content for
users based on their past interactions.
n:

• Robotic Process Automation (RPA): AI

automates repetitive tasks in industries like
Process finance and manufacturing. E.g.,
Autom Automating invoice processing.
ation:
 Popularity: Python is widely used due to its
simplicity and readability.

Ease of Use: Python’s syntax is easy to learn,

making it accessible for beginners.

Why Python for AI Libraries:

and ML? • NumPy: For numerical computations.
• pandas: For data manipulation and analysis.
• scikit-learn: For ML algorithms.
• TensorFlow and PyTorch: For deep learning.

Example: Python script using pandas to load a

dataset and scikit-learn to train a model.
 Data Types: Integers, floats, strings, lists,
dictionaries.
• Example: x = 5 (integer), y = 3.14 (float), name =
"Alice" (string).

Variables: Assigning values to variables.

Basic Python • Example: a = 10; b = "Hello"
Concepts
Basic Syntax: Example of a simple Python
script.
• Example:

def greet(name):
return f"Hello, {name}!“
print(greet("World"))
 Types of Data:

• Structured Data: Organized in rows and columns

(e.g., databases).
• Example: An Excel sheet with sales data.
• Unstructured Data: Not organized in a pre-defined
manner (e.g., text, images).
• Example: Tweets, customer reviews.
What is Data?
Importance of Data: Data is the foundation
of AI and ML; it drives model training and
predictions.

Example: Dataset of housing prices with

features like size, location, and price.
 Data Cleaning:
Removing Example: Handling
inconsistencies, missing values by
handling missing values, filling them with the
standardizing data mean or median.
formats.

Data Preparation: Example: Scaling

Data Processing Normalizing data,
feature engineering.
features to have zero
mean and unit
variance.

Data Visualization:
Tools like Matplotlib Example: A
and Seaborn for histogram showing
visualizing data the distribution of
distributions and housing prices.
relationships.
 Role in AI and ML:
Essential for understanding
data distributions, making
predictions, and validating
models.

Probability Distributions: • Example: Bell curve

Why Statistics and Normal distribution, binomial

distribution, etc.
representing the
normal distribution.

Probability? Events and Outcomes:

• Example: Flipping a
coin and calculating
Basic concepts in probability. the probability of
heads or tails.
 Mean: Average value. Example:
Mean of [1, 2, 3, 4, 5] is 3.

Median: Middle value. Example:

Measures of Central Tendency:
Median of [1, 2, 3, 4, 5] is 3.

Key Statistical Mode: Most frequent value.

Concepts Example: Mode of [1, 2, 2, 3, 4] is 2.

Variance: Measure of data spread.

Example: Variance of [1, 2, 3, 4, 5]
is 2.

Measures of Dispersion:

Standard Deviation: Square root of

variance. Example: Standard
deviation of [1, 2, 3, 4, 5] is 1.41.
 Supervised Learning: Learning from labeled data.
• Classification: Predicting categories (e.g., spam detection).
• Example: Email spam filter.
• Regression: Predicting continuous values (e.g., house prices).
• Example: Predicting housing prices based on size and location.

Unsupervised Learning: Learning from unlabeled

Types of Machine data.

Learning • Clustering: Grouping similar data points (e.g., customer

segmentation).
• Example: Grouping customers based on purchasing behavior.
• Dimensionality Reduction: Reducing the number of features
(e.g., PCA).
• Example: Reducing the dimensions of image data.

Reinforcement Learning: Learning by interacting

with the environment (e.g., game playing).
• Example: Training an AI to play chess.
 • Example: Collecting • Example: Handling
sales data from missing values,
different stores. normalizing data.

Data Collection: Data

Gathering data Preprocessing:
from various Cleaning and
sources. preparing data.

ML Workflow
Model Model Training:
Evaluation: Using algorithms
Assessing model to learn from
performance. data.
• Example: Using • Example: Training a
metrics like accuracy, linear regression model
precision, recall to to predict sales.
evaluate a classification
model.
 Linear Regression: Example: Predicting
Predicting continuous values house prices based on
based on input features. square footage and
location.

Decision Trees: Splitting Example: Classifying

data into branches to make whether a customer will
predictions. buy a product based on age
and income.

K-Means Clustering:
ML Algorithms
Example: Segmenting
Grouping data points into customers into groups
clusters. based on purchasing
behavior.

Random Forest: An Example: Predicting loan

ensemble of decision trees to defaults by combining
improve accuracy. multiple decision tree
outputs.

Support Vector Machines Example: Classifying

(SVM): Classifying data emails as spam or non-
points by finding the optimal spam.
hyperplane.
 Metrics:

• Accuracy: Proportion of correct predictions.

• Example: Accuracy of a spam filter (e.g., 95% of emails
correctly classified).
• Precision: Proportion of true positive predictions.
• Example: Precision of 90% means 90% of emails
classified as spam are actually spam.
• Recall: Proportion of actual positives correctly identified.
Model Evaluation • Example: Recall of 85% means 85% of actual spam
emails were correctly identified as spam.
• F1 Score: Harmonic mean of precision and recall.
• Example: Balancing precision and recall in a
classification task.
• Cross-Validation: Technique for assessing model
performance by dividing data into training and validation
sets.
• Example: 10-fold cross-validation to evaluate the model’s
performance on different subsets of data
 • Explainable AI: Efforts to make
AI decisions understandable by
humans.
• Example: Models that provide
human-readable explanations
for their predictions.
• AI in IoT: Integration of AI with
Internet of Things devices.
• Example: Smart home devices
that learn user preferences over
Future Trends in AI time.
• Edge Computing: Processing
and ML AI on local devices rather than
centralized servers.
• Example: AI-enabled cameras
that process video data locally
for faster decisions.
• Advancements in AI Research:
Mention recent breakthroughs
and future possibilities.
• Example: GPT-4 and other
large language models
transforming natural language
processing.
 Simple example of a Random Forest sales prediction model using Python
and the Scikit-learn library:

• Let's assume we have a dataset with the following features:

Random Forest • product_id: Unique identifier for each product
Classification • price: Price of the product
• ad_spend: Amount spent on advertising for the product
• season: Season in which the product was sold (e.g., summer, winter)
• sales: Target variable, total sales for each product