AI Glossary
The complete dictionary of Artificial Intelligence
Deep Neural Networks
Architecture of artificial neural networks with multiple hidden layers that enables learning of complex hierarchical data representations.
Adaptive Complex Systems
Collections of interconnected agents whose emergent behavior results from local interactions and which dynamically adapt to their environment.
Non-Linear Interactions
Relationships between system components where the output is not proportional to the input, creating unpredictable and emergent behaviors.
Forward Propagation
Computation process in a neural network where information flows from the input layer to the output layer through synaptic weights.
Non-Linear Activation Function
Mathematical function applied to each neuron to introduce non-linearity, allowing the network to learn complex relationships.
Cascade Architecture
Neural network structure where layers are connected sequentially with skip connections to facilitate deep learning.
Recursive Neural Networks
Networks applied to tree or graph data structures, recursively processing nodes to capture complex hierarchies.
Chaotic Dynamic Systems
Systems whose temporal evolution exhibits extreme sensitivity to initial conditions, making their long-term behavior unpredictable.
Multi-Scale Modeling
An approach consisting of representing a complex system at different temporal and spatial scales to capture its intrinsic dynamics.
Attention Mechanisms
Technique allowing the network to selectively weigh different parts of the input to focus on the most relevant information.
Memory Neural Networks
Architecture integrating external memory components to store and retrieve information, enabling reasoning over long sequences.
Emergent Properties
Complex behaviors or characteristics that appear at the system level but are not present in its individual components.
Information Theory in Networks
Application of information theory principles to analyze and optimize the flow of information in deep neural networks.
System-Network Coupling
Bidirectional integration between a complex physical system and a neural network for real-time prediction and control.