AI Terminology Part 1

L1 Terminology (AI) Part 1

• Artificial intelligence (AI) encompasses the engineering of intelligent machines and programs.
• Machine learning (ML) is a subset of AI, enabling computers to learn without explicit programming.
• Deep learning (DL) is a specific type of machine learning that uses neural networks. Neural networks are composed of interconnected nodes or layers that process and learn from data.
• AI aims to accomplish tasks such as problem-solving, recognition (e.g., face, voice), and decision-making, sometimes improving upon human capability.
• Voice assistants (Siri, Alexa, Google Assistant) exemplify AI. They work by understanding questions, searching for answers, and providing responses.
• AI relies on several layers to accomplish what may seem like simple operations.

Relationship Between AI, ML, and DL

• AI emerged around the 1950s.
• Machine Learning gained traction in the 1980s.
• Deep learning emerged in 2010.
• Deep learning is a subset of machine learning, and machine learning is a subset of AI.

Artificial Intelligence (AI)

• AI is a branch of computer science focused on creating systems capable of performing tasks requiring learning, reasoning, and problem-solving.
• AI serves to assist humans with tasks such as problem-solving, recognition (examples include facial and vocal recognition), and decisions. Sometimes, AI surpasses human capabilities in these areas.

Artificial intelligence (AI) - Example: Voice Assistants

• Voice assistants function by understanding questions, finding answers, and formulating responses.
• The process involves multiple layers of AI capabilities.

Machine Learning (ML)

• Machine learning is a field that provides computers with the ability to learn without explicit programming.
• A prominent pioneer in machine learning is Arthur Samuel. He defined machine learning as a field which enables computers to learn without being explicitly programmed.
• Machine learning is a means to achieve artificial intelligence through systems or software models which learn from experience to identify patterns in data.

Deep Learning (DL)

• Deep learning is a subset of machine learning. It employs neural networks to detect complex patterns in data.
• Neural networks are structured in layers of interconnected nodes which learn from data.
• Deep learning's ability to extract complex patterns comes from multiple layers of processing and learning from data.

Deep Learning (DL) - Identifying a person from thousands of images of faces

• Deep learning models, trained on multiple images, identify unique characteristics of a face, such as angle or facial expressions.
• Different conditions such as lighting and angles do not hinder the ability of the model to identify a person.
• Applications are seen across various areas, such as securing phones with face ID and identifying people using security protocols.

Relationship Between AI, ML, and DL

• AI emerged in the 1950s.
• ML developed in the 1980s.
• DL emerged in the 2010s.
• Deep learning is a subset of machine learning, which is a subset of artificial intelligence.

Tools and Platforms for AI Implementation

• Tools exist to facilitate AI implementation.
• These tools encompass pattern recognition, pattern detection, data, prediction, model inference, and model pruning.
• High-performance, advanced CPUs and GPUs are essential tools for tackling complex algorithms within AI.
• Cloud computing platforms offer scalable resources for AI applications.
• This allows researchers and developers to train models without heavy initial investments.

Why Use Cloud Computing for AI?

• Cloud computing offers computational resources without needing substantial hardware.
• The high-performance nature of servers in cloud computing settings aids in higher processing speeds than traditional computing systems.
• Cloud providers are competitors, resulting in cost effectiveness among alternative providers.

Which Programming Language is Best for AI?

• AI can be developed with diverse programming languages, including Python, C++, Java, and others.
• Python stands out for its simple syntax, along with compatibility to modules and packages, making it attractive for rapid application development tasks.

What is Python?

• Python was created by Guido van Rossum and released in 1991.
• Python is widely used for its rapid application development capabilities due to its simple and user-friendly syntax.
• Python benefits from a high degree of flexibility with existing modules and packages.

Python as an Interpreted Language

• Python is an interpreted language.
• Programming code needs to be compiled or interpreted before execution.
• Python code is converted to binary code for execution.
• Python code executes line-by-line, in contrast with compiled languages.

Python Installation with Anaconda

• Anaconda is a free and open-source Python distribution.
• It is a common method for installing Python, particularly for Windows, macOS and Linux.

Why Use Virtual Environments for Different Projects?

• Virtual environments are essential to avoid conflicts between different project requirements that may need different library versions.
• Using a separate environment for each project ensures consistency and avoids conflicts between project requirements that demand distinct library versions.
• This approach enhances project management efficiency and enables faster development processes.

Conda channel

• A conda channel serves as a repository for packages, which conda utilizes for installing packages.
• Channels can include multiple packages and are accessed via Conda commands.
• The default conda channel offers stability, simple management, and a limited package collection.

Conda-forge

• Conda-forge is a community-driven, accessible channel for Conda.
• It provides a variety of packages.
• Key features include: community maintenance, cross-platform compatibility, and enhanced dependency resolution compared to the default channel.

PyPl

• PyPI is the official repository for Python packages. Developers can share and publish their libraries in this repository.
• PyPI is a comprehensive repository of Python libraries. It offers a wide selection of packages.
• The most recent versions of packages are often found in PyPl.

Introduction to Programming Basics

• Python programming fundamentals include variables, data types, conditional statements, loops, and functions.

What Are Variables and Data Types in Programming?

• Variables are used to store values in programming languages.
• Common data types include integers (int), floating-point numbers (float), strings (str), and booleans (bool).

Using Variables and Data Types

• Python uses dynamic typing, making variable declaration streamlined.
• In Python, the declaration of variables and data types are done automatically by Python's interpreter.

What Are Conditions in Programming?

• Conditions are crucial for decision-making within programs.
• A specified code block is executed only if the condition evaluates to true.

Using Conditions

• Conditions are useful for making decisions within code.
• Code blocks are executed based on the evaluation of conditions.

What Are Loops in Programming?

• Loops enable repeated actions without redundant code.
• The for loop is a prevalent tool for programmers.

Using Loops

• Utilizing loops leads to more concise and efficient code for repeating tasks.
• Avoiding repetitive code makes the code clearer and easier to scale.

What Are Functions in Programming?

• Functions are blocks of code designed to perform specific tasks.
• Defining functions utilizes the def keyword, followed by the function name, arguments, and a colon (:).
• Functions receive input data, process it, and potentially return a result.

Using Functions

• Functions enable streamlined data processing.
• Functions are fundamental for organizing and streamlining code.

When Do We Use Libraries?

• Libraries, composed of modules (Python files), are utilized when specific, complex functionality is required.
• Instead of writing all code from scratch, pre-built functions make the process more efficient.

What Are Libraries in Python

• Libraries are collections of modules that offer specific Python functionalities.
• Libraries are often used to avoid reinventing the wheel by providing pre-built, tested, and optimized functionality for diverse applications.

Popular Python Libraries

• Popular libraries encompass: Math, Random, Numpy for calculations; Pandas for data manipulation; Matplotlib for data visualization; and Scikit-learn for machine learning.
• These libraries save effort by providing ready-to-use functions.

Calculation Library: Math

• The math library is a crucial library for performing mathematical calculations within Python.
• It facilitates the usage of the sqrt function to extract the square root of a selected number.

Calculation Library: Random

• The random library generates random numbers, frequently implemented in simulations and games.
• The randint function generates random integers within specified ranges.

Calculation Library: Numpy

• The NumPy library supports numerical computations within Python.
• NumPy is adept at managing data arrays. It offers a broad selection of high-level mathematical functions.
• NumPy is ideal for working with multi-dimensional arrays and matrices effectively.

Data Science Library: Pandas

• The Pandas library excels for data manipulation.
• Pandas is suitable for data analysis where data is structured in tabular format, such as data from a CSV file.
• Pandas is highly effective in managing tabular data.

Plotting Library : Matplotlib

• Matplotlib is a powerful plotting library for producing high-quality 2D figures, such as line charts, bar charts, histograms, and more.
• Matplotlib aids in data visualization for analysis and decision-making.

Machine Learning Library: Scikit-learn

• Scikit-learn is an advanced Python machine learning library.
• It's capable of handling smaller and larger datasets, enabling the creation of machine learning models for practical applications.
• Scikit-learn encompasses various machine learning algorithms.

Machine Learning Process

• Machine learning algorithms are trained using data examples, referred to as training sets.
• A larger training set generally results in more accurate models.
• Each item in a training set is classified using either 0 or 1 as a labeling scheme.
• Training data, including images, videos, and other data types, serve as inputs for machine learning algorithms.

Find Data: Collect Data Ourselves

• Data collection can be done hands-on using sensors (e.g., measuring humidity or electrical activity in healthcare applications).

Find Data: Datasets

• Datasets are pre-existing archives of data which may be difficult to collect or very costly to source.
• A well-known example of an expensive sensor is LiDAR.
• Complex data collection may be cost prohibitive. Cancer data, for instance, may prove difficult to collect.

Where to Download Datasets for Al Projects

• Kaggle (https://www.kaggle.com/) supplies a wealth of datasets.
• The UC Irvine Machine Learning Repository (https://archive.ics.uci.edu/) is another valuable resource.
• ImageNet (https://image-net.org/) offers diverse image datasets and other online sources.
• Other options exist such as TensorFlow (https://knowyourdata.withgoogle.com/).

Datasets Type: Tabular

• Tabular data is presented in a spreadsheet or CSV format of rows and columns.
• Common applications of this data include tasks such as classification and regression.

Datasets Type: Image

• Image datasets are vital for computer vision, stored as JPEGs or PNGs.
• Image datasets are used in tasks like facial recognition and object detection.

Datasets Type: Text and Audio

• Text datasets (.txt) are data collections for general use.
• Audio datasets (e.g., in WAV or MP3 format) are available for diverse, widespread use cases.

Google Utilizing AI

• Google applications utilize AI for various functions, from photo organization and email filtering to real-world tasks like scheduling appointments.
• Examples include smart search on Google photos, smart compose and reply features on Gmail, and AI assistant capabilities in Google Assistant.
• Data from various sources is trained by machine learning algorithms to identify patterns and produce relevant outputs.

This quiz covers the foundational terminology related to Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL). You'll explore the definitions, historical developments, and examples of these innovative technologies, including popular voice assistants. Test your knowledge and understanding of these crucial AI concepts!