AI-ordlista
Den kompletta ordlistan över AI
Spiking Neuron Models
Mathematical representations of biological neurons including Leaky Integrate-and-Fire, Hodgkin-Huxley and Izhikevich to simulate the temporal dynamics of action potentials.
Temporal Coding and Encoding Schemes
Methods for converting continuous data into spike trains including rate coding, time coding, rank coding and latency coding to represent information in SNNs.
Synaptic Plasticity and STDP
Bio-inspired learning rules where synaptic connection strength evolves based on temporal delays between pre- and post-synaptic spikes.
Supervised Learning Algorithms
Training techniques including SpikeProp, backpropagation through time, and gradient-based methods adapted for discontinuous spiking networks.
Deep SNN Architectures
Multilayer spiking network structures including convolutional, recurrent, and attentional SNNs for complex recognition tasks.
Neuromorphic Hardware and Spiking Chips
Specialized integrated circuits such as Loihi, TrueNorth, and SpiNNaker designed to efficiently execute spiking computations with low energy consumption.
SNN for Computer Vision
Application of spiking networks to image and video processing with event-based sensors (DVS) for real-time and energy-efficient perception.
SNN for Signal Processing
Use of spiking networks for audio analysis, speech recognition and temporal signal processing exploiting their intrinsically temporal nature.
Spiking Neural Network Simulation and Frameworks
Specialized software tools such as Brian2, NEST, BindsNET, and SpyTorch for modeling, simulating, and training spiking neural networks.
Spiking Information Theory
Theoretical study of coding capacity, energy efficiency and fundamental limits of spiking networks based on information theory.
Hybrid SNNs and Conversion
Methods for converting traditional artificial neural networks into SNNs and hybrid architectures combining spiking and non-spiking neurons.
Spiking Neural Reservoirs
Fixed random recurrent networks with spiking neurons used as computational reservoirs for time series processing with simple output learning.