Please enable JavaScript.
Coggle requires JavaScript to display documents.
Introduction for AI - Coggle Diagram
Introduction for AI
Disciplines (Fields connected to AI)
Linguistics (language understanding)
Machine learning
Philosophy
Robotics
Neural networks
Control theory & Cybernetics
What is AI
Tasks that require human brains, like learning, problem-solving, and language comprehension, are automated by AI. AI poses problems like job shifts and ethical quandaries in sectors like banking, healthcare, and transportation.
Acting like humans
Organizing and Setting Up
Logic with Fuzzy
Knowledgeable Systems
In-depth Education
Thinking rationally:
Decision Theory:
Argumentation via Abduction
Theory of Games
Inductive reasoning
Acting rationally:
Active
Social Media
Flexible
Objective-focused
Thinking like humans:
The study of cognition
State of the Art (Latest AI Technologies)
Robots and Smart devices
Face recognition Ex: Apple Face ID
Game playing Ex: AlphaGo by DeepMind
Speech Recognition Ex: Siri, Google Assistant, Alexa
Chinese Room
Computers can provide answers by adhering to rules without fully comprehending them, as shown by the Chinese Room thought experiment.
self awareness
sapience
subjectivity
perception
History of AI
Early Foundations (1950s–1960s)
Turing Test (1950):
A tool that assesses a machine's capacity for human-like intelligence was created by Alan Turing.
Perceptron (1958):
An early neural network model for simple pattern recognition was presented by Frank Rosenblatt.
MADALINE (1960):
One of the first neural networks to be applied to real-world problems.
Hopfield Network & XOR Problem (1969):
While XOR demonstrated the limitations of early neural nets in single-layer perceptrons, Hopfield networks supplied memory.
Neural Network Advances (1980s–1990s)
Backpropagation (1986):
Deep learning for multi-layer networks is made possible by a novel algorithm.
Q-Learning & CNN (1989):
While image identification mainly relied on Convolutional Neural Networks (CNNs), Q-learning introduced reward learning.
Support Vector Machines (SVM) (1995):
An effective machine learning technique for resolving classification issues is presented in this text.
LSTM (1997):
Sequential data learning, a critical component of speech and text processing, has been successfully resolved by long short-term memory networks.
Deep Learning Era (2000s–2016)
First Deep Learner (1992–1993):
The individual is actively involved in conducting deep architecture training experiments.
ImageNet (2009):
A huge image collection made with sophisticated visual AI.
ReLU (2011):
A novel activation function was successfully used to train deep networks.
Dropout (2012):
A technique for lessening overfitting in deep learning is presented in this text.
GANs & DeepFace (2014):
DeepFace improved facial recognition, and Generative Adversarial Networks enabled the creation of images.
AlphaGo (2016):
In the game of Go, DeepMind AI defeated a human champion, demonstrating strategic thinking.
Web Links
https://www.simplilearn.com/tutorials/artificial-intelligence-tutorial/what-is-artificial-intelligence
https://cloud.google.com/learn/what-is-artificial-intelligence
https://www.bitlaw.com/ai/intro-to-AI.html
Types of AI
Weak AI
Common applications such as voice assistants and recommendation systems often use this term.
concentrates on a single task.
The individual adheres to predetermined guidelines rather than truly "thinking."
Strong AI
The idea has not yet been fully developed and is still theoretical.
The objective is to build robots with human-like understanding and reasoning.
Like people, machines are capable of learning, adapting, and problem-solving.