Deep Learning (DL)

These terms cover the essential concepts and architectures that are foundational to deep learning, highlighting the structures and methodologies used to build models capable of learning from data in ways that mimic the depth and complexity of human learning processes.

Activation Function - Functions that decide whether a neuron should be activated or not, influencing the neuron's output based on its input.

Artificial Neural Network (ANN) - The foundational structure for deep learning, composed of layers of interconnected nodes or "neurons" that process data.

Autoencoder - A type of neural network used for unsupervised tasks like dimensionality reduction or feature learning, where the network is trained to encode input data as representations and then decode these representations back to the original input format.

Convolutional Neural Network (CNN) - A class of deep neural networks, most commonly applied to analyzing visual imagery, using convolutional layers to process data in a grid-like topology.

Deep Learning (DL) - A subset of machine learning involving neural networks with many layers (deep networks) that learn representations of data through a hierarchy of increasing complexity or abstraction.

Hidden Layer - Layers within a neural network that are neither input nor output layers. In deep learning, networks often contain multiple hidden layers, contributing to the "depth" of the network.

Input Layer - The first layer in a neural network that receives the input signal to be processed by subsequent layers.

Large Language Model (LLM) - A type of deep learning model that processes and generates natural language, trained on vast amounts of text data. Examples include models like GPT (Generative Pretrained Transformer).

Loss Function - A function that measures the difference between the actual output of the model and the expected output, guiding the training of deep learning models by minimizing this loss.

Neuron - A basic unit of computation in neural networks, modeled after biological neurons, which processes inputs and generates an output signal.

Output Layer - The layer in a neural network that produces the final output of the model, providing the end result of the computation.

Recurrent Neural Network (RNN) - A type of neural network where connections between units form a directed cycle, allowing it to use internal memory to process sequences of inputs, useful in time-series data or natural language.

Transformer Architecture - A model architecture designed for handling sequential data, without requiring the sequence to be processed in order. It relies heavily on self-attention mechanisms and is widely used in state-of-the-art natural language processing tasks.