Grade 11 Student Notes: AI Foundations
**Unit 2: AI Applications and Methodologies**
**Summary:**
- Introduction to positive AI applications across science, healthcare, business, and weather forecasting.
- Understanding cognitive computing concepts like Computer Vision and speech recognition.
- Recognition of AI as a supplement to human capabilities.
- Insight into future societal changes due to AI and required adaptation skills.
**Key Concepts:**
AI applications, cognitive computing, Impact of AI on society.
**Activities:**
- Explore societal perceptions of AI through online image search.
- Evaluate online articles on AI for relevance and accuracy.
**AI Applications: Chatbots**
- Definition: AI applications simulating conversations with users.
- Examples: HDFC Bank's EVA, IBM Watson Assistant.
- Real-life applications: Customer service, E-commerce, Healthcare.
**Types of Chatbots:**
1. Rule-based Chatbot: Follows predefined rules for simple conversations.
2. AI-based Chatbot: Capable of holding complex conversations and learning from experience.
**Natural Language Processing (NLP):**
- Definition: Technology enabling machines to understand and process human language.
- Objective: Interpret, comprehend, and generate meaningful responses in human language.
1. Weather Prediction using AI:
Tools for Weather Forecasting:
1. Utilizes satellite data, mathematical models, and AI algorithms for accurate predictions.
2. Leading tools include IBM GRAF and Panasonic's TAMDAR.
Types of Weather Forecasting Techniques:
1. Traditional methods compare historical data with current observations.
2. AI-based techniques use computer vision and machine learning for real-time adjustments and higher
accuracy.
2. Price Forecast for Commodities:
Commodities like agriculture products are prone to price fluctuations.
Transition from classical mathematical models to AI-based systems offers better accuracy and micro-level
predictions.
3. Self-Driving Cars:
Utilize sensors like radar, lidar, and GPS combined with AI algorithms for navigation.
Enhance safety and efficiency in transportation, reducing accidents and improving traffic flow.
Characteristics and Types of AI:
Characteristics include autonomy, predictive capabilities, continuous learning, and adaptability.
Types include data-driven AI, autonomous systems, recommendation systems, and human-like AI (AGI).
4. Cognitive Computing:
Built on AI and signal processing, mimicking human brain functions like perception, learning, and reasoning.
Applications include healthcare, where AI aids in diagnosis and treatment decisions.
5. AI and Society:
Positive impacts include advancements in healthcare, transportation, and disaster prediction.
Challenges include biased data, technological unemployment, and privacy concerns.
6. Understanding Deep Learning:
Deep learning mimics human decision-making processes using neural networks.
Applications range from image recognition to natural language processing.
Supervised Machine Learning Activity:
Utilizes teachable machine to understand AI components and train classifiers.
Testing different datasets helps in understanding the role of training data in AI systems.
Notes:
Natural Language Processing (NLP) is a branch of artificial intelligence (AI) that focuses on enabling computers to
understand, interpret, and generate human language. It has found applications across various fields, including text
recognition, summarization, information extraction, and speech processing.
Text Recognition:
NLP is utilized for text recognition in images or videos, such as reading vehicle number plates.
Cameras or machines capture images, which are then processed by NLP algorithms to extract relevant text data.
Quality of image plays a crucial role in accurate data extraction.
Summarization by NLP:
NLP can summarize lengthy documents or articles by extracting key phrases or shortening the source text.
Two methods of summarization are extraction-based (using key phrases) and abstraction-based (shortening the
document).
Applications include legal document analysis, news summarization, and research paper abstract generation.
Information Extraction:
NLP aids in extracting structured information from unstructured sources, such as emails or documents.
It identifies entities, relationships, and attributes within text data to provide meaningful insights.
Common applications include search engines and automated report generation.
Speech Processing:
Speech processing involves computers analyzing and understanding human speech content.
Devices like virtual assistants (e.g., Alexa, Siri) utilize NLP to interpret user queries and provide relevant responses.
Microphones capture sound waves, which are converted into frequency graphs and then analyzed by NLP algorithms.
Notes: Set B Topics
Semantic Segmentation
Definition: Image classification process in Computer Vision.
Example: Identifying objects in an image and outputting a class or probability.
Complexity: Natural for humans, complex for machines.
Classification and Localization
Definition: Identifying and bounding objects in images.
Example: Predicting object class and creating bounding boxes.
Importance: Enhances understanding of object positions in images.
Object Detection
Description: Identification of multiple objects in images or videos.
Example: Algorithm identifies and localizes various objects.
Application: Used for diverse purposes like surveillance and automation.
Instance Segmentation
Explanation: Technique outlining distinct objects in images.
Method: Creates pixel-wise masks for individual objects.
Benefit: Offers granular understanding of objects in images.
Weather Prediction using AI
Case Studies: Fani cyclone impact, Indian Meteorological Department failures.
Video: Impacts of weather forecasting on society.
Tools: IBM GRAF, Panasonic's TAMDAR for weather forecasting.
Price Forecast for Commodities
Definition: Prediction of commodity prices using AI.
Significance: Affects farmers, policymakers, and industries.
Shift to AI: Benefits include higher accuracy and micro-level forecasting.
Self-Driving Car
Description: Autonomous vehicles using AI for navigation.
Features: Sensors, deep learning, robotics, and navigation systems.
Applications: Transportation, safety, and efficiency improvements.
Characteristics and Types of AI
AI Features: Autonomy, prediction, continuous learning, and reactivity.
Types: Data-Driven AI, Autonomous Systems, Recommendation Systems, and Human-Like AI.
Data-Driven AI
Definition: AI systems trained on large datasets.
Advantages: Predictive capabilities, learning from vast data, real-time adjustments.
Autonomous System
Explanation: Technology operating without human intervention.
Examples: Self-driving cars, Mars rovers, smart home systems.
Functionality: Utilizes AI for decision-making and task execution.
Recommendation Systems
Function: Suggests products or information based on user data.
Examples: YouTube video recommendations, Flipkart product suggestions.
Importance: Enhances user experience and engagement.
Human-Like AI
Definition: AI systems simulating human-like intelligence.
Examples: Jarvis from "Iron Man," Artificial General Intelligence (AGI).
Challenges: Current AI not achieving true human-like intelligence.
Cognitive Computing
Description: AI platform utilizing machine learning and natural language processing.
Applications: Medical treatment support, decision-making assistance.
Advantages: Mimics human brain functions, aids in complex decision-making.
Applications of Cognitive Computing
Uses: Medical treatment, disaster prediction, transportation, agriculture.
Case Study: IBM Watson for Oncology at Memorial Sloan Kettering Cancer Centre.
AI and Society
Benefits: Healthcare improvements, transportation innovations.
Challenges: Integrity of AI, technological unemployment, privacy concerns.
Impacts: Positive and negative effects on various aspects of society.
Non-Technical Explanation of Deep Learning
Concept: Deep learning mimics human decision-making processes.
Analogy: Delivery route optimization to understand neural networks.
Complexity: Deep neural networks with multiple layers for complex tasks.
Image Datasets
Three different datasets include:
Dataset Description
Initial
Training
Dataset
These are the images students should use to “teach” their machine learning
model which image is a cat and which image is a dog.
Note that there are many more cats and that the cats are more diverse in
appearance than the dogs. This means that the classifier will more accurately
classify cats than dogs.
Test Dataset These are the images that students should use to test their classifier after training.
Students should show these images to their model and record if their classifier
predicts if the image is of a dog or a cat.
Note: Students should not use these images to teach their classifier. If an image is
used to train a classifier, the machine will have already recorded the
corresponding label for the particular image. Showing this image to the machine
during the testing phase will not measure how well the model generalizes.
Recurating
dataset
This is a large assortment of images students can use to make their training
dataset of cats and dogs larger and more diverse.
The test dataset should be used twice, once for testing students’ initial classifier and again for testing their
recurated dataset
