Machine Learning

Described as the science that enables machines to analyze, translate, and interpret data to solve complex problems.  The three types of ML are Supervised Learning, Unsupervised Learning, and Reinforcement Learning.

Supervised:  Labeled data is used find patterns and correlations between data. The input and the output data are defined.

Unsupervised:  Unlabeled data is used to help build data relationships, find patterns, and identify correlations within the data set. 

Reinforcement: Rules are defined within programs, and the machine executes a task based on the results of the established rules.

A quote from Jason Brownlee describes ML like this:  “Machine Learning is getting computers to program themselves. If programming is automation, then machine learning is automating the process of automation.” 

Neural Network

Neural Networks are systems designed to learn how to perform specific tasks by analyzing data and then executing those tasks without limitations of existing rules. They are networks that are designed to operate in a similar manner to our own biological signal-sending brain. These networks are also called Artificial Neural Networks (ANNs). 

A system of nodes exist and the use of inputs and outputs determine if (and how) data is sent to the next layer. Programmers develop powerful algorithms to assist with training these networks and improving the accuracy of their results.

Repetitive training and algorithm tuning results the machine’s ability to rapidly process and classify data as quickly as the data can proceed through the layers.  Internet search engines such as Google are excellent examples of a finely tuned neural network in action.

Natural Language Processing

Natural Language Processing (NLP) is used in communications between computer systems and humans and is considered to be a type of machine translation. Within AI, NLP helps computers interpret human language and bridges gaps between our diverse languages and how a computer analyzes that language in both speech and text.

NLP is important because it allows computers to translate words in a consistent manner.  Systems using NLP can process massive amounts of data without growing tired or making errors due to fatigue.

NLP within AI is critical because within Machine Learning (supervised and unsupervised) there will always be an element of human dialect, grammar rules, and misused terminology. NLP helps analyze those variances in text and voice and translates them into machine language that a computer can understand.  Thank you Spell Check and Alexa.

Expert

Systems

Expert Systems replicate the decision-making process and intelligence of expert humans. Samudyata Bhat provides a great explanation: “Breaking down an expert system essentially is an AI software that uses knowledge stored in a knowledge base to solve problems.”

Systems designed like this rely heavily on information and knowledge stored within the system for constant analysis based on user searches and requests.

Processes executed within Expert Systems typically utilize if/then statements or rules, comprised of two parts: “If” is the precondition, and “Then” is the consequence.

Expert Systems such as robust search engines and document-creating/editing software are designed to be reliable, consistent, responsive, and thorough.

One limitation of expert systems that is worth of mentioning is that they lack creativity and cannot replicate our ability to develop a unique solution.