One stop for learning most of the important algorithms for building your Artificial Intelligence Application. This repository contains coded examples of all the algorithms that will help you learn not just the theory of how the algotithm works but also the implementation of it in python programming language.
Artificial Intelligence is about making systems that ACT RATIONALLY. We build Agents that perceive the environment through sensors and process the information to act rationally.
Until the 1970's humans used to try to mimic the human brain, but did not turn out to be much successful. The lessons learnt from the brain were -
Memory and Simulation are key to decision making.
An agent is an entity that perceives and acts accordingly. A rational agent selects actions that maximize its expected utility. Characteristics of the percepts, environment and action space dictate techniques for selecting rational actions.
There are two types of agents -
-
Reflex agents
Reflex agents are Agents that dont plan at all. They just take a decision about their next action based on their current state. They do not consider the future consequences of their actions and so are not Rational. They just consider how the world is now and do not have a model of the world.
-
Planning agents
Planning agents ask the question "What if?", and this is what makes its decision intelligent. Its decisions are based on hypothesized consequences of actions. These agents consider how the world would be, if the take this action. They need a model of how the environment/world would evolve in response to their actions. There are two types of these planning agents -
-
Complete planning agents
These agents compute their complete path before starting execution. These agents can take up a long time before making their first move, as they look through all possible states of the world. The path these agent find is the optimal one, but the process of planning can be slow.
-
Re-planning agents
These agents plan on each step they take. These are faster and are guaranteed to find the goal but the solution may not be optimal.
-
There are a lot of new terms that were introduced in the previous definition. I will go through all of them, and as we go on and start implementing the algorithms, you will understand these concepts even better.
Environment is nothing but the world for which the agent was created. An example for this can be, the game environment. Imagine you were building an agent to play the game of pacman. The environment includes of everything in the pacman world, the dots, walls, ghosts, power-ups and the pacman itself.
Action space is the possible set of actions the agent can take. Like in the previous example of pacman, the action space was (UP, DOWN, LEFT, RIGHT). The agent will be the pacman itself and the action space is the possible actions the agent can select.
Like action space, state space is the possible set of states the world can be in. With the same example of pacman, each position of the agent could correspond to a different game state and a set of all possible such states is what corresponds as State Space. Some special kind of states are goal state, start state and current stat.
Hope this gives you a good introduction to the world of AI. You will get a better idea these concepts as the course progresses. Hope you have a great time doing this course and I am looking forward to see the intelligent application that you build learning the basics from this course.