Monday, September 4, 2023

Demystifying Artificial Intelligence: Categories and Subcategories

1. Basic AI: Narrow or Weak AI (ANI)

SubCategory_1 : Chatebots

  • Chatbots represent rule-based conversational agents, designed to engage with users, answer inquiries, and offer assistance within predefined conversational boundaries.💬
  • Example: - Customer service chatbots deployed on e-commerce websites.
  • Tool:-  Dialogflow by Google facilitates the creation of conversational agents with minimal coding.

      SubCategory_2 : Virtual Assistants

      • Virtual assistants are AI-driven entities capable of executing tasks, offering information, and responding to user queries, often through natural language understanding and generation.🤖
      • Example:- Siri and Google Assistant provide intelligent assistance and information retrieval. 
      • Tool:- Developers can integrate virtual assistants using Apple's SiriKit and Google's Assistant SDK.

        2. Generative AI: Machine Learning (ML)

        SubCategory_1 : Supervised Learning
        • Supervised learning involves training machine learning models on labeled datasets to enable predictions and classifications based on pattern recognition. . 📈
        • Example: Email spam filters leverage supervised learning to identify and filter out spam messages.
        • Tool: Scikit-Learn, a Python library, streamlines the development and training of supervised ML models

        • Problem:  Predicting Student Grades
          • Scenario: Imagine you're a teacher, and you want to predict the final grades of your students based on the number of hours they studied for an exam. You have collected data from previous students, including the number of hours they studied and their actual exam grades.
          • Data : Here's a simplified dataset with a few examples:
                                    Hours Studied (X) Exam Grade (Y)
                                            1                            60
                                            2                            70
                                            3                            75
                                            4                            85
                                            5                            90
          • Objective:The objective is to build a supervised machine learning model that can predict a student's exam grade (Y) based on the number of hours they study (X).

        SubCategory_ 2 : Unsupervised Learning

        • Unsupervised learning encompasses techniques for extracting patterns and structures from unlabeled data, including clustering and dimensionality reduction.📊
        • Example: Unsupervised learning may be applied to group customers based on purchase behavior.
        • Tool: The K-means clustering algorithm in scikit-learn is a popular choice for clustering tasks.
        • Problem: Clustering Grocery Store Customers
          • Scenario:Imagine you're the manager of a grocery store, and you want to understand your customers better for marketing purposes. You have a dataset that includes purchase history but no specific labels or categories for customers.
          • Data: Here's a simplified dataset with a few examples:
                                  Customer ID Amount Spent (in dollars)    Items Purchased
                                                1                     100                                            10
                                                2                       50                                              5     
                                                3                     200                                            20
                                                4                     150                                            15
                                                5                     300                                            30
          • Objective:The objective is to use unsupervised learning to group or cluster customers based on their purchase behavior, without any prior information about customer categories

        SubCategory_ 3: Reinforcement Learning

        • Reinforcement learning employs trial-and-error methods guided by reward mechanisms to enable agents to learn sequences of actions for optimal decision-making.🎮
        • Example: Reinforcement learning is prevalent in training game-playing AI agents to maximize rewards.
        • Tool: OpenAI Gym provides a suite of environments for experimentation with reinforcement learning.
        • Problem: Navigating the FrozenLake
          • Scenario: Imagine you're controlling an agent (e.g., a robot or a person) trying to traverse a frozen lake. The goal is to reach the destination (a goal cell) while avoiding holes in the ice. The agent starts at a designated starting point and can take actions to move in four directions: up, down, left, or right. The agent receives a reward for each step and a higher reward for reaching the goal. However, falling into a hole results in a negative reward.
          • Environment: The FrozenLake environment can be represented as a grid world:

                                    S: Starting point
                                    F: Frozen surface (safe to walk)
                                    H: Hole (dangerous)
                                    G: Goal
          • Objective:The objective for the RL agent is to learn a policy (a strategy) that guides it to reach the goal while avoiding the holes, maximizing the cumulative reward

          3. Super Smart AI: Deep Learning (DL)

          SubCategory_1: Convolutional Neural Networks (CNNs)

          • CNNs are deep learning models specialized in hierarchical feature extraction from structured data, particularly well-suited for image analysis and recognition.🖼️
          • Example: CNNs excel in facial recognition and object detection tasks.
          • Tools: TensorFlow and Keras offer robust platforms for developing and training CNNs.
          • Problem: Handwritten Digit Recognition
            • Scenario: The task is to build a CNN that can recognize and classify handwritten digits (0-9) from grayscale images.
            • Data: We'll use the MNIST dataset, which contains 28x28 pixel grayscale images of handwritten digits (0-9). Each image is labeled with the corresponding digit.
            • ObjectiveThe objective is to train a CNN to correctly classify these digits based on the image data.

          SubCategory_2: Recurrent Neural Networks (RNNs)

          • RNNs are designed for handling sequential data by incorporating feedback connections, making them suitable for tasks like natural language processing and time-series analysis.📜
          • Example: RNNs predict the next word in a sentence, making them valuable in language modeling.
          • Tool: PyTorch is a versatile framework for implementing RNNs in various applications.
          • Problem: Text Generation
            • Scenario: The task is to build an RNN that can generate text, character by character, based on a given input text. We'll train the model on some example text and then use it to generate new text in a similar style.
            • Data: For simplicity, we'll use a short example text:  "Hello, how are you doing today?"
            • Objective: The objective is to train an RNN to predict the next character in the text sequence based on the previous characters. Once trained, the model can be used to generate new tax

          SubCategory_3: Generative Adversarial Networks (GANs)

          • GANs employ a generative and discriminative network in adversarial collaboration to produce high-quality synthetic data or content.🎨
          • Example: GANs can create realistic images from textual descriptions, contributing to image synthesis.
          • Tool: TensorFlow and PyTorch provide comprehensive support for GAN development.
          • Problem: Generating Synthetic Handwritten Digits
            • Scenario: The task is to create a GAN that can generate new images of handwritten digits that look similar to those found in the MNIST dataset.
            • Data: We'll use the MNIST dataset for training. It consists of 28x28 pixel grayscale images of handwritten digits (0-9).
            • Objective: The GAN consists of two neural networks: a Generator and a Discriminator. The Generator tries to create realistic images, while the Discriminator tries to distinguish between real images from the MNIST dataset and fake images generated by the Generator. The Generator and Discriminator play a game, where the Generator improves its ability to create realistic images, and the Discriminator improves its ability to tell real from fake

            4.Language Understanding: Natural Language Processing (NLP)

            SubCategory_1: Sentiment Analysis

            • Sentiment analysis involves the automatic assessment of emotions, opinions, and sentiments expressed in textual data. 😃😢
            • Example: Sentiment analysis is instrumental in evaluating public sentiment through social media text.
            • Tool: NLTK (Natural Language Toolkit) is a foundational library for sentiment analysis tasks.
            • Problem: Sentiment Analysis on Movie Reviews
              • Scenario:vThe task is to perform sentiment analysis on movie reviews to classify them as positive or negative based on the text content.
              • Data:vWe'll use a small dataset containing movie reviews, where each review is labeled as either positive or negative.
              • Objective:vThe objective is to build a sentiment analysis model that can predict the sentiment of a given movie review text as either positive or negative.

            SubCategory_2: Named Entity Recognition (NER)

            • NER is a subtask of NLP focused on identifying and categorizing named entities, such as names of individuals, locations, or organizations, in text.🏙️👤
            • Example: NER extracts and categorizes names and locations from news articles.
            • Tool: spaCy, a Python library, is known for its efficiency in NER tasks.

              5. Seeing the World: Computer Vision

              SubCategory_1: Object Detection

              • Object detection AI identifies and localizes objects or entities within images or videos.📷🔍
              • Example: Object detection is fundamental in identifying and tracking vehicles in traffic camera footage.
              • Tool: YOLO (You Only Look Once) is a renowned framework for real-time object detection.

              SubCategory_2: Facial Recognition

              • Facial recognition AI involves the automated identification of individuals based on facial features and patterns.👤🔓
              • Example: Facial recognition technology is utilized in unlocking devices through facial authentication.
              • Tool: OpenFace is a leading solution for facial recognition applications.

                6. Doing Real Things: Robotics

                SubCategory_1: Autonomous Vehicles

                • Autonomous vehicles merge AI and robotics to navigate and control vehicles without human intervention, relying on sensors and decision-making algorithms. 🚗🤖
                • Example: Autonomous cars autonomously navigate city streets and highways.
                • Tool: ROS (Robot Operating System) is widely adopted for developing autonomous vehicle systems.

                SubCategory_2: Home Automation

                • Home automation robots assist in household tasks, such as cleaning, with the aim of improving convenience and efficiency.
                • Example: The Roomba robot vacuum autonomously cleans floors within homes.🏡🤖
                • Tool: iRobot Create 2 serves as a platform for DIY robotics projects in home automation..

                You may also like

                Kubernetes Microservices
                Python AI/ML
                Spring Framework Spring Boot
                Core Java Java Coding Question
                Maven AWS