Guiding Backprop by Inserting Rules
نویسندگان
چکیده
We report on an experiment where we inserted symbolic rules into a neural network during the training process. This was done to guide the learning and to help escape local minima. The rules are constructed by analysing the errors made by the network after training. This process can be repeated, which allows to improve the network performance again and again. We propose a general framework and provide a proof of concept of the usefullness of our approach.
منابع مشابه
The Obturator Guiding Technique in Percutaneous Endoscopic Lumbar Discectomy
In conventional percutaneous disc surgery, introducing instruments into disc space starts by inserting a guide needle into the triangular working zone. However, landing the guide needle tip on the annular window is a challenging step in endoscopic discectomy. Surgeons tend to repeat the needling procedure to reach an optimal position on the annular target. Obturator guiding technique is a modif...
متن کاملLearning Many Related Tasks at the Same Time with Backpropagation
Hinton [6] proposed that generalization in artificial neural nets should improve if nets learn to represent the domain's underlying regularities . Abu-Mustafa's hints work [1] shows that the outputs of a backprop net can be used as inputs through which domainspecific information can be given to the net . We extend these ideas by showing that a backprop net learning many related tasks at the sam...
متن کاملBackprop as Functor: A compositional perspective on supervised learning
A supervised learning algorithm searches over a set of functions A → B parametrisedby a spaceP to find the best approximation to some ideal function f : A → B. It does this by taking examples (a, f (a)) ∈ A×B, andupdating the parameter according to some rule. We define a categorywhere these update rules may be composed, and show that gradient descent—with respect to a fixed step size and an err...
متن کامل2 MECHANISMS OF MULTITASK BACKPROPWe
Hinton 6] proposed that generalization in artiicial neural nets should improve if nets learn to represent the domain's underlying regularities. Abu-Mustafa's hints work 1] shows that the outputs of a backprop net can be used as inputs through which domain-speciic information can be given to the net. We extend these ideas by showing that a backprop net learning many related tasks at the same tim...
متن کاملCarotid-compression technique for the insertion of guiding catheters.
Inserting a guiding catheter into a tortuous artery for neurointerventional procedures can be difficult. In our technique, the carotid artery is manually compressed to stabilize and/or straighten the inserted wire before advancing the guiding catheter. Although this technique is not risk-free and care must be taken to avoid vascular injury by excessive compression, it is useful for the insertio...
متن کامل