Major Subfields of AI - Robotics

 Robotics is a major subfield of AI that deals with designing, constructing, and operating robots to carry out different jobs without or with minimal human intervention. It integrates AI with engineering to develop smart machines that can perceive, think, and engage with their environment. It also includes computer vision, machine learning, and control systems.

Fundamental Elements of Robotics

• Creating the various mechanical components of a robot, from the design to the assembly of the actuators and the sensors, is fundamental to the construction and design of a robot.
• Robots apply various tools and techniques, such as sensors and computer vision, to perceive and understand the world around them, which is essential for navigation and recognizing objects.
• Robots must improve their performance using statistical learning. Autonomous decision-making constitutes a complex cognitive skill that can be learned through trial and error.
• Robots must be able to execute a sequence of tasks and control their movement and balance through software and algorithm control which allows them to automate control performance and execute a sequence of tasks.
• Human-robot systems can use natural language processes for voice command recognition and other techniques for interaction to provide intuitive control and collaboration of a robot with a human.
• Robots can be programmed to create paths that link two locations and to navigate complex environments through a process of selection.

Major Subfields of AI - Natural Language Processing (NLP)

 Natural Language Processing (NLP) is a subdiscipline of AI that aims at allowing computers to comprehend and manipulate human language. The most important ones are Natural Language Understanding (NLU) and Natural Language Generation (NLG), which processes and generates language respectively. The fundamental operations include text analysis by use of tasks such as tokenizing and part-of-speech tagging and the applications are broadly used in areas such as translation, sentiment analysis and chatbots.

Core componentsNatural Language Understanding (NLU): The process of enabling a computer to derive meaning from human language, often involving mapping input to a useful representation.
Natural Language Generation (NLG): The process of producing meaningful human-like text from data or an internal representation.

Common tasks and techniques

• Text Preprocessing:Tokenization: Breaking text into smaller units like words or sentences.
• Part-of-Speech (POS) Tagging: Assigning a grammatical category (e.g., noun, verb) to each word.
• Named Entity Recognition (NER): Identifying and categorizing entities such as people, organizations, and locations.
• Syntax and Parsing: Analyzing the grammatical structure of a sentence.
• Semantic Analysis: Understanding the meaning of words in a given context.
• Discourse and Pragmatic Analysis: Analyzing the relationships between sentences and understanding context.

Key applications

• Machine Translation: Translating text from one language to another.
• Sentiment Analysis: Determining the emotional tone or opinion expressed in text, often used for social media and customer feedback.
• Chatbots: Creating conversational interfaces that can understand and respond to user input.
• Text Summarization: Automatically creating a concise summary of a larger document.
• Information Extraction: Pulling specific, useful data from unstructured text.

Major Subfields of AI - Machine Learning (ML)

Machine learning (ML) represents an important sub-field of AI where algorithms are utilized to learn when presented with data, and its key sub-types are supervised, unsupervised, semi-supervised and reinforcement learning. Additional essential fields in or connected to the field of ML are deep learning, based on neural networks; self-supervised learning and transfer learning. 

Key subfields of ML

• Supervised Learning: This is where algorithms are trained on labelled data to make a prediction.
• Unsupervised Learning: It is a training process which learns to identify patterns and structures using data which is not labeled.
• Semi-Supervised Learning: Labelled and unlabelled data are combined to acquire an understanding.
• Reinforcement Learning: These algorithms learn on a trial and error basis being rewarded or punished by their actions.
• Deep Learning: A sub-field of ML based on deep neural networks inspired by the human brain to learn based on large datasets.
• Self-Supervised Learning: This is a form of supervised learning, except that the data is also self-labeled.
• Transfer Learning: The model that has been trained on one task is re-used as the initial point of a model on a second task. 

Applications of ML

• Image recognition and speech recognition.
• Predictive analytics
• Recommendation systems
• Natural language processing (NLP)?
• Computer vision