Glossario IA
Il dizionario completo dell'Intelligenza Artificiale
Population Decoding
Technique for interpreting collective activity patterns from a group of neurons to reconstruct the original temporal information of the processed audio signal.
Spike Latency
Temporal delay between input stimulation and spike generation, used as an information code in SNNs to represent the temporal characteristics of speech signals.
Leaky Integrate-and-Fire (LIF) Neuron
Basic neuronal model simulating the gradual leakage of membrane potential and generating spikes when the activation threshold is reached, suitable for real-time processing of audio streams.
Synaptic Filter
Mechanism for temporal weighting of incoming spikes based on synaptic dynamics, enabling selective extraction of frequency features in audio signals.
Coincidence Detection
Ability of SNNs to identify synchronized temporal events between different inputs, fundamental for spatial localization of sound sources and source separation.
Event-Based Processing
Computational paradigm where neurons process information only upon the occurrence of spikes, optimizing energy consumption for continuous processing of audio streams.
Neuronal Synchronization
Temporal alignment of spikes between neuronal populations, used to segment and group temporal components of complex speech signals.
Adaptive Activation Threshold
Variable spike triggering threshold that adjusts according to neuronal activity history, enabling dynamic sensitivity to variations in sound levels.
Short-Term Spiking Memory
Ability of SNNs to temporarily maintain temporal information through membrane potential states, essential for processing vocal sequences.
Temporal Source Separation
Application of SNNs exploiting temporal pattern differences to isolate individual components in complex audio mixtures.
Temporal Pattern Recognition
Process of identifying specific spike sequences corresponding to pre-defined acoustic patterns, used for word and phoneme recognition.
Spiking Wavelet Transform
Neural implementation of time-frequency analysis using spikes to represent wavelet coefficients, adapted for real-time analysis of non-stationary signals.