Lecture 2 & 3: Machine Learning and Large Language Models (LLMs)

The Era of Big Data

• The widespread use of various digital systems leads to the era of “big data”

• We are both producers and consumers of data

• Data is informative

• We need “big theory” to convert data to knowledge

Machine Learning

• Machine learning is programming computers to optimize a performance criterion using example data or past experience

  • Learning general models from a data of particular examples

• Advantages:

  • Ability to review large volumes of data and identify patterns and trends that might not be apparent to a human

  • Improves the accuracy over time

  • No need for human intervention (automation)

Categories of Machine Learning

Classification

• Classification is the task of assigning objects to one of several predefined categories

• Given a collection of records (training set)

• Each record is by characterized by a tuple (x,y), where x is the attribute set and y is the class label

• Task: learn a model that maps each attribute set x into one of the predefined class labels y

Regression

• Regression analysis is a statistical method that helps us to analyze and understand the relationship between two or more variables of interest

• The process that is adapted to perform regression analysis helps to understand which factors are important, which factors can be ignored, and how they are influencing each other

• Task: learn a model that maps each attribute set into a continuous value

Regression vs. Classification

Clustering

• Clustering is finding groups of objects such that the objects in a group will be similar (or related) to one another and different from (or unrelated to) the objects in other groups

• Examples Spam filters, marketing and sales, and identifying fraudulent or criminal activities

Unsupervised Outlier Detection

• An outlier is an observation which deviates so much from the observations as to arouse suspicions that it was generated by a different mechanism

• Semantically close to audit terminology such as audit exceptions and transaction anomalies

• Often contains useful information about abnormal characteristics of the systems and entities

  • Abnormal debit/credit amount, occurrence time, counterparty

  • Possible reasons: mechanical faults, changes in system behavior, fraudulent behavior, human error, etc.

Reinforcement Learning

• Reinforcement learning is a machine learning training method based on rewarding desired behaviors and/or punishing undesired ones

• In general, a reinforcement learning agent is able to perceive and interpret its environment, take actions and learn through trial and error (e.g., resource management)

Machine Learning Improve Expert Systems

• Expert system (rule-based): rather than attempting to identify new patterns and classify data based on statistics or operational experience, they are designed to replicate the problem-solving process of experts in the field

  • Always based on known rules

  • A direct application of existing knowledge

  • Do not attempt to learn directly from the data

• Rule-based machine learning (RBML) is a term in computer science intended to encompass any machine learning method that identifies, learns, or evolves 'rules' to store, manipulate or apply

Comments on Adopting Machine Learning

• Computers are able to create the system

  • Learning from the data

• The key to success of machine learning applications:

  • Enough data

  • Less noisy

  • Limited time and money

• Traditional methods and machine learning can benefit each other

Large Language Models (LLMs)

• Large language models (LLMs) are a type of artificial intelligence model designed to understand and generate human-like text

• They are based on a type of neural network called a transformer, which uses self-attention mechanisms to understand the context and meaning of words in a sentence

• A form of task-specific AI, but is highly versatile in natural language processing tasks

• Not true AGI but exhibits signs of “General understanding” within the realm of language

• Leverages vast amounts of text data to generate coherent and contextually relevant responses

• Impressive but limited: lacks consciousness, reasoning like humans, or broad adaptability outside of language

The Development of GPT Models

GPT-2 & Training Controversy

• GPT-2, introduced in 2019, was a more sophisticated version with 1.5 billion parameters

• It was capable of generating longer and more coherent text

• The release of GPT-2 sparked controversy due to concerns about potential misuse for generating fake news and misleading content

GPT-3 & GPT-3.5 Turbo (ChatGPT)

• GPT-3, introduced in 2020, was the most advanced model yet, with 175 billion parameters.

• It was trained on an enormous corpus of text data, including books, articles, and web pages

• GPT-3.5 Turbo (ChatGPT) is a conversational AI model that can engage in human-like conversations, trained on a large corpus of conversational data

GPT-4 & Multimodal Training

• GPT-4, introduced in 2023, is a multimodal large language model capable of understanding more than text, such as images

• It was pretrained to predict the next token using both public data and data licensed from third-party providers, and then fine-tuned with reinforcement learning from human and AI feedback

Task-Specific AI vs. General AI

• Task-Specific AI (Narrow or Weak AI)

  • Designed for one specific task

  • Examples: Siri, facial recognition software

  • Cannot perform tasks outside its programmed domain

  • Dominates current AI applications

• General AI (Artificial General Intelligence, AGI)

  • Mimics human cognitive functions

  • Can perform any intellectual task a human being can

  • Adaptable to unfamiliar tasks without specific training

  • Still theoretical; not yet fully realized