A
-
Agent
In AI, an agent is an entity that perceives its environment through sensors and acts on it through actuators to achieve specific goals. Agents can be simple (such as a thermostat) or complex (such as a virtual assistant that learns and adapts).
-
Algorithm
An algorithm is a set of defined instructions or rules that a machine follows to solve a problem or perform a specific task. In AI, algorithms are essential for processing data, learning from it and making decisions.
-
Algorithmic bias
A phenomenon in which an algorithm produces systematically biased or unfair results due to faulty training data, design assumptions or limitations in the model. It can reinforce existing inequalities or create new forms of discrimination in AI applications.
B
-
Backpropagation
(Backward propagation) Algorithm used in the training of artificial neural networks. It adjusts the weights of the network by calculating the output error and propagating it backwards through the layers, updating the weights to minimize the error. It is key for networks to learn efficiently.
-
Bias
(Bias) In neural networks, bias is an additional value introduced into each neuron that allows the activation function to be shifted to the left or right. This helps the model to better fit the data and improve its learning capability. In a broader sense in AI, bias can also refer to unwanted biases or tendencies in the data or models.
C
-
Chatbot
An artificial intelligence program designed to simulate a human conversation, either by text or voice. Chatbots can answer questions, perform tasks or hold natural dialogues, using simple rules or advanced natural language processing (NLP) models.
-
Cognitive Computing
A branch of artificial intelligence that seeks to mimic the functioning of the human brain. It combines natural language processing, machine learning and automated reasoning to interpret complex data, make decisions and learn from experience.
D
-
Deep Learning
Subfield of machine learning based on multi-layer artificial neural networks (deep neural networks). These networks are capable of learning complex representations and performing advanced tasks such as speech recognition, computer vision or natural language processing.
-
Data Augmentation
A technique used to expand training data sets by creating modified versions of existing data. This helps to improve the generalizability of AI models, especially in contexts where the amount of original data is limited.
-
Data Mining (Minería de Datos)
Extract valuable information from large data sets, like panning for gold in a mine.
E
-
Explainable AI (XAI)
Approach that seeks to make AI systems understandable to humans.
F
-
Fine-tuning
Fine tuning of a pre-trained model for specific tasks.
-
Framework
Development environment for building and training AI models (such as TensorFlow or PyTorch).
G
-
Generative AI
Artificial intelligence that creates original content such as text, images or music.
-
GPT (Generative Pre-trained Transformer)
Language model capable of generating coherent text.
H
-
Heuristics
Strategy that simplifies decision making in AI based on approximate rules.
-
Hyperparameter
External variable affecting model training (such as learning rate).
I
-
Inference
Use of a trained model to make predictions or generate answers.
J
-
JSON (JavaScript Object Notation):
Lightweight data format used to transfer information in IA.
-
Jupyter Notebook
Interactive environment for programming, documenting and visualizing AI models in Python.
K
-
K-Means
A clustering algorithm that organizes data into clusters or groups.
L
-
LLM (Large Language Model)
Large-scale language model trained with large volumes of text.
M
-
Machine Learning
A branch of artificial intelligence that allows machines to learn from data and improve their performance on specific tasks without being explicitly programmed for each situation. It uses algorithms that identify patterns and make predictions or classifications.
N
-
Neural Network (Red Neuronal):
A computational model inspired by the human brain, made up of interconnected neurons that process information.
Ñ
Nothing here yet
O
-
Overfitting:
When a model fits too closely to training data and does not generalize well to new data. It is like memorizing instead of understanding.
P
-
Pattern Recognition:
Area of artificial intelligence that focuses on identifying patterns and regularities in data. It uses machine learning and statistical techniques to recognize structures, classify information and make predictions based on previous examples.
-
Prompt
An instruction given to a language model to generate a response.
Q
-
Q-Learning:
Reinforcement learning algorithm that teaches an agent to make decisions.
R
Nothing here yet
S
-
Supervised Learning
Supervised learning, where the data includes the correct answer.
T
-
Token
Minimum unit (word, syllable, character) with which language models work.
-
Transformer
Key neural network architecture in models such as GPT, BERT, or Claude.
-
Training
The process by which a model learns patterns from data.
-
TensorFlow
A software tool for creating AI applications.
U
-
Unsupervised Learning
Unsupervised learning, the model finds patterns without labeled data.
-
Underfitting (Subajuste)
When a model is too simple to capture the complexity of the data.
V
-
Validation
Evaluate the performance of an AI model on an independent dataset to ensure its effectiveness.
W
-
Web scraping
Technique for automatically extracting data from websites, useful for feeding AI models.
X
-
XAI (Explainable Artificial Intelligence)
Explainable and transparent artificial intelligence.
Y
-
YAML
Configuration language used in model training and AI deployment.
Z
-
Zero-shot Learning
The ability of a model to generalize to tasks it has never seen during training.