AI Dictionary




AI Factory: 









In the context of machine learning, explainability refers to the ability to understand and interpret how a model makes predictions or decisions. It involves making the internal workings and decision-making processes of a complex model transparent and understandable to humans. Explainable AI (XAI) is particularly important in situations where the decisions made by a model impact individuals or society, as it helps build trust, accountability, and aids in identifying potential biases or errors.Various techniques and tools are employed to enhance the explainability of machine learning models, such as feature importance analysis, model-agnostic interpretability methods, and generating human-understandable explanations for specific predictions. Achieving explainability is crucial for ensuring that machine learning systems are used responsibly and ethically, especially in sensitive domains like healthcare, finance, and criminal justice.

Ensemble Methods

Ensemble methods are techniques that aim at improving the accuracy of results in models by combining multiple models instead of using a single model. The combined models increase the accuracy of the results significantly. Used in both statistics and machine learning, ensemble methods use multiple learning algorithms to obtain better predictive performance than could be obtained from any of the constituent learning algorithms alone



Generative Adversarial Networks (GANs)

GANs are a class of machine learning models designed for unsupervised learning. They consist of two neural networks—the generator and the discriminator—engaged in a competitive interaction. The generator creates synthetic data, attempting to mimic real examples, while the discriminator evaluates these samples, distinguishing between real and generated data. This adversarial process drives the generator to continuously improve its output, creating increasingly realistic data. GANs find applications in generating images, videos, and other content, showcasing their ability to produce high-quality, authentic-looking results.








Mixture of experts (MoE) is a machine learning technique where multiple expert networks (learners) are used to divide a problem space into homogeneous regions. It differs from ensemble techniques in that typically only one or a few expert models will be run, rather than combining results from all models.


Neural Network



Parameter Efficient Fine-Tuning (PEFT) offers an effective solution by reducing the number of fine-tuning parameters and memory usage while achieving comparable performance to full fine-tuning. The demands for fine-tuning PLMs, especially LLMs, have led to a surge in the development of PEFT methods. So PEFT is a form of instruction fine-tuning that is much more efficient than full fine-tuning - with comparable evaluation results as you will see soon. PEFT is a generic term that includes Low-Rank Adaptation (LoRA) and prompt tuning (which is NOT THE SAME as prompt engineering!).



Retrieval-Augmented Generation (RAG)

Retrieval-Augmented Generation or RAG refers to the process of retrieving relevant information from external knowledge bases, before then answering questions with LLMs. RAG has been demonstrated to significantly enhance answer accuracy, reduce model hallucination, particularly for knowledge-intensive tasks. The RAG AI framework is thus able to utilise the most accurate, up-to-date information and to give users insight into LLMs' generative process.

Reinforcement learning

Reinforcement learning is a type of machine learning that operates without explicit supervision, enabling algorithms to acquire skills through a process of trial and error. The fundamental concept involves using a system of rewards and punishments, often likened to a "carrot and stick" approach. In this framework, when an algorithm attempts to enhance a task, it receives a positive "reward" for success or "punishment" for failure, say based on feedback. Through repeated iterations, the algorithm improves its performance, sometimes exceeding human capabilities, especially if the learning environment closely mirrors the complexities of the real world.




Supervised Learning

Supervised learning is a machine learning paradigm where the algorithm is trained on a labelled dataset, meaning it is provided with input-output pairs during training. The goal is for the algorithm to learn the mapping from inputs to corresponding outputs, allowing it to make predictions or classifications on new, unseen data. This type of learning involves a supervisor or teacher guiding the algorithm by providing explicit feedback, helping it to generalize and make accurate predictions on new, unseen data. Supervised learning is commonly used in tasks such as image recognition, speech recognition, and regression problems.


SAP, or Systems, Applications, and Products in Data Processing, is an enterprise software suite that facilitates various business processes. It integrates functions like finance, human resources, and logistics, enabling organizations to streamline operations, enhance efficiency, and make informed decisions. SAP systems are widely used for managing complex business operations and data, offering a comprehensive platform for businesses to achieve integration and optimization across different departments.


Training data

Training data refers to the subset of data used to train a machine learning model. It consists of input-output pairs, where the input represents the features or attributes of the data, and the output is the corresponding label or target. The purpose of training data is to enable the machine learning algorithm to learn the underlying patterns and relationships within the data, allowing the model to make accurate predictions or classifications on new, unseen data. The quality and representativeness of the training data significantly impact the performance and generalization ability of the trained model. Choosing diverse and well-labelled training data is crucial for developing robust and effective machine learning models.


Unsupervised Learning: