Abstract
How networks endure damage is a central issue in neural network research. In this paper, we study the slow and fast dynamics of network damage and compare the results for two simple but very different models of recurrent and feed forward neural network. What we find is that a slower degree of network damage leads to a better chance of recovery in both types of network architecture. This is in accord with many experimental findings on the damage inflicted by strokes and by slowly growing tumors. Here, based on simulation results, we explain the seemingly paradoxical observation that disability caused by lesions, affecting large portions of tissue, may be less severe than the disability caused by smaller lesions, depending on the speed of lesion growth.
Highlights
The performance of networks enduring damage is a central issue in biological and non-biological networks (Cohen et al, 2000b; Nawrocki and Voyles, 2011)
To be more realistic η should be a function of the network disability
We define a modified learning rate as a tool for better interpretation of the results of dynamic changes based on the network damage and disability
Summary
The performance of networks enduring damage is a central issue in biological and non-biological networks (Cohen et al, 2000b; Nawrocki and Voyles, 2011). Recovery from a lesion with given size is possible depending on whether the lesion evolved slowly or fast (Duffau et al, 2002, 2003; Desmurget et al, 2007; Varona, 2010) This is important because in a realistic setting of optimal recovery, the temporal rate as well as the size of injury must be taken into account. The biological basis of recovery after stroke and how the brain reorganizes itself, for example, is still largely unknown (Pantano et al, 2003; Calautti and Baron, 2003) but certainly it is possible that the degree and speed of recovery varies considerably for different lesion locations and depends on structural alterations taking place in the spared brain tissue during a slow and gradual temporal lesion (Stewart and Ades, 1951; Meyer et al, 1958; Glick and Zimmerberg, 1972)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
