How that somewhat unconstrained structure of an artiﬁcial neural network copes with generalization, especially when there are several ‘competiting’ stimuli, that simultaneously want to be ‘extrapolated’ onto ‘regions’ in the inputs space of the FNN not covered by the training data.
On Interference of Signals and Generalization in Feedforward Neural Networks
This brings in the beautiful structure of its group of isometries and certain of its subgroups which have a direct interpretation in terms of the organization of the neural populations that are assumed to encode the structure tensor.
Hyperbolic planforms in relation to visual edges and textures perception
These examples are a few among many of an analysis that would have important implications in terms of the actual neural representation of the structure tensor (and at bottom of the image intensity derivatives).
Hyperbolic planforms in relation to visual edges and textures perception
In computer science and engineering acyclic or approximately acyclic graphs occur in data structures, software call graphs, and feed-forward neural networks.
Random graph models for directed acyclic networks
Remote Sensing, 14. Schalkoff, R., 1992. Pattern Recognition: Statistical, Structural, and Neural Approaches.
Fish recognition based on the combination between robust feature selection, image segmentation and geometrical parameter techniques using Artificial Neural Network and Decision Tree
Scale–free networks are associated with a variety of structures and systems, such as protein folding and biopolymer dynamics; cell metabolism; neural, information and communication networks; various evolutional, ecological, social and economical systems.
On the motifs distribution in random hierarchical networks
T. Y. Kwok and D. Y. Yeung, “Constructive algorithms for structured learning in feedforward neural networks for regression problems,” IEEE Transactions on Neural Networks, vol. 8, 1997, pp. 630-645. M.
RGANN: An Efficient Algorithm to Extract Rules from ANNs
It is hard to envisage a model at the level of neural networks which successfully represent and communicate its own global informational structures.
Definability in the Real Universe
Our mean ﬁeld approach may be extended to rewiring processes starting from other than ring-like structures, e.g. to two or three dimensions, as for instance relevant for neural network modeling .
Small-world spectra in mean field theory
Our approach allows for a more ﬂexible and structured way of dealing with the temporal aspects in the neural code.
Dynamic State Estimation Based on Poisson Spike Trains: Towards a Theory of Optimal Encoding
To sum up, the aim of the thesis is to shed light on how cellular dynamics can lead to the complex network structures of neural systems, and, in its turn, in what ways this topology can inﬂuence, optimise and determine the collective behaviour of such systems.
Interplay between Network Topology and Dynamics in Neural Systems
Development of neural network structure with biological mechanisms, S.
Interplay between Network Topology and Dynamics in Neural Systems
Structured information in small-world neural networks.
Interplay between Network Topology and Dynamics in Neural Systems
Development of neural network structure with biological mechanisms.
Interplay between Network Topology and Dynamics in Neural Systems
Stein, Thermodynamic chaos and the structure of short-range spin glasses, in: Mathematical aspects of spin glasses and neural networks, 243-287, Progr.
(Non-) Gibbsianness and phase transitions in random lattice spin models
***