When the average human being thinks of Artificial Intelligence(AI), usually the thoughts that come to mind are not very positive. As many people believe it is a replacement of human jobs such as cashiers, truck drivers, taxi drivers and, construction workers. There seems to be a stigma of AI when there is an involvement of the topic within a conversation. What is rarely taken into consideration within most people is that there as so much more that is involved in AI and so many capabilities in hand. It is not here to replace jobs but moreover help humans thrive more and in general help within their daily lives to be more efficient.
So, what is AI, what does it do, and what is its purpose?
There are many different definitions of AI and what it means to humans, but the most common ones are
1. A branch of computer science dealing with the stimulation of intelligent behavior in computers
2. The capability of a machine to imitate intelligent human behavior
3. Simulation of human intelligence processes by machines, these processes include learning, reasoning, and self-correction
Now to sum all of these definitions, Artificial Intelligence is essentially the ability of a computer program or a machine to think and learn.
Trending AI Articles:
1. Basics of Neural Network
2. Bursting the Jargon bubbles — Deep Learning
3. How Can We Improve the Quality of Our Data?
4. Machine Learning using Logistic Regression in Python with Code
It can be classified into two categories
Weak AI: Is a form of AI focused on one narrow task and is never taken as general intelligence but rather a construct designed to be intelligent in a narrow task assigned to it. A good example is Apple’s Siri which has the internet behind it serving as a robust database. Though it seems very smart, it operates in a very narrow, predefined manner. It can be found when it has inaccurate results as sometimes it is unable to engage in some conversations as it is not programmed to respond.
Strong AI: More similar to a human and can think and perform tasks on its own. It is capable of all and any cognitive functions that a human has.
There are numerous ways where AI can be achieved. Some ways AI can be created through machine learning is vision, Reinforcement learning, and Natural Language Processing (NLP).
Arthur Samuel back in 1959 stated his definition: “[Machine Learning is the] field of study that gives computers the ability to learn without being explicitly programmed.”
Four types of machine learning within Artificial Intelligence:
1. Supervised Learning
2. Unsupervised learning
3. Semi-supervised learning
4. Reinforcement learning
Supervised learning: Is the machine learning task of learning function that maps the input to be paired to an output. The majority of practical machine learning uses supervised learning. Supervised learning is where you have input variables (x) and an output variable (Y) and you use an algorithm to learn the mapping function from the input to the output Y = f(X). There are two ways to predict the data set in supervised learning.
Classification: Is a predictive modeling task of approximating a mapping function (f) from the input variable (x) to discrete output variable (y). The output variables are often called labels or categories. For a given observation, the mapping function predicts the class or category. It can have real-valued or discrete variables.
Example of Classification would be when a primary task is to create an algorithm to classify whether an image contains a dog or a cat. The input for this task is images of dogs or cats from the training dataset, while the output is the classification accuracy on the test dataset. For the classification to occur upon a data set, it needs to look for specific details to classify weather is something is a cat or a dog. Appearances between the two can be that a cat has a long tail and whiskers whereas a dog does not have whiskers and has a short tail.
Regression: predictive modeling which is a task of approximating a mapping. function (f) from input variables (x) to a continuous output variable (y). The continuous output variable is a real-value that is a usual an integer/floating-point value. Linear regression is a parametric method, which means it assumes the form of the function relating to X and Y.
An Example could be predicting income. The input data X includes all relevant information about individuals in the data set that can be used to predict income, such as years of education, years of work experience, or job title. These attributes are called features, which can be numerical or categorical. In the regression model below, the years of post-secondary(x), can be used to predict the income(y) an individual can perhaps, obtain.
Unsupervised learning: The data given to the unsupervised algorithm are not labeled which means only the input variables(x) are given with no corresponding output variables. Unsupervised learning can be further grouped into clustering and association problems.
Clustering: Is a type of problem where similar things are grouped. A clustering problem is where you want to discover the inherent groupings in the data, such as grouping customers by purchasing behavior.
An example can be how scientists want to divide different species up based on their characteristics(x), to have a more clear view of which animals have similar characteristics. Below is the data set that represents animals using the variables distance and bearing. There is a before and after use of the clustering learning method for unsupervised learning.
Association: It is a rule-based machine learning method for discovering interesting relations between variables in large databases.
Semi-supervised learning: algorithms develop mathematical models from incomplete training data, where a portion of the sample inputs is missing the desired output. Problems where you have a large amount of input data (X) and only some of the data is labeled (Y) are called semi-supervised learning problems.
Reinforcement learning (RL): Is a type of machine learning technique that enables an agent to learn in an interactive environment by trial and error using feedback from its own actions and experiences. The goal is to find a suitable action model that would maximize the total cumulative reward of the agent. The figure below represents the basic idea and elements involved in a reinforcement learning model. Within the image, Environment is a physical world in which the agent operates, State is a current situation of the agent, Reward is Feedback from the environment, Policy is a method to map the agent’s state to actions, and lastly Value is Future reward that an agent would receive by taking an action in a particular state.
Computer Vision: This field of learning enables machines to see. Machine vision captures and analyses visual information using a camera, analog-to-digital conversion, and digital signal processing. It seeks to automate tasks that a human visual system can do. It is interlinked usually with machine learning in order to get the best results.
An example of where computer vision is being utilized currently is in the Amazon go store. As they have cameras to detect when a shopper picks up and puts down an item. Amazon has developed cameras and sensors that can recognize individuals, track them around the store, know which account is linked to each customer, understand exactly which product and how many of each are put in your bag.
Natural Language Processing (NLP): Is the ability of a computer program to understand human language as it is spoken. It can be used to interpret a free text and make it analyzable. It is based upon deep learning and uses patterns in data to improve a program’s understanding.
Functionalities of Artificial Intelligence:
Artificial Intelligence can be classified into four types when it comes to functionality.
Type 1 (Reactive Machines): This is the most basic type that is unable to form memories and use past experiences to inform decisions. They are unable to perform tasks outside of specific tasks they were designed for. An example is Deep Blue, the IBM chess program that beat Garry Kasparov in the 1990s. This machine can identify pieces on the chess board and make predictions, even though is had no memory and cannot use past experiences to inform future ones. Its sole priority was to choose the most strategic moves and was designed for narrow purposes as it had no other job other than to play chess.
Type 2(Limited Memory): Has the ability to use past experiences to inform future decisions. It depends on pre-programmed knowledge and observations carried out over time. Some of the decision-making functions are designed in self-driving cars. Observations inform actions happening in the not-so-distant future, such as a car making a turn or changing lanes.
Type 3(Theory of Mind): This term refers to the understanding that others have their own personal beliefs, desires, and intentions that impact the decisions they make.
Type 4 (Self-awareness): Artificial Intelligence has gained the qualities of a human, they have a sense of self and have a consciousness. Machines with self-awareness understand their current state and can use that information to infer what others may feel. This is not existent and is categorized as Strong AI.
- Artificial Intelligence has the capabilities of doing much more than what it is intended for currently
- Machine learning is a type of AI that can teach forms of AI without being explicitly programmed
- Weak AI has the capability to only perform one task
- Strong AI has the ability to mimic how a human act but has very little presence in the current world
- There are different functionalities of AI dependent on what tasks are needed to be performed