Abu-Mostafa YS. Hints and the VC dimension. Neural Comput 1993;5(2):278--88. Home/Search Document Not in Database Summary Related Articles Check |
This paper is cited in the following contexts:
Neural Processing Letters 10: 231--242, 1999. - Knowledge Incorporation Into (Correct)
....if several related tasks are learned at the same time. Hence, the second main approach to the incorporation of knowledge into neural networks is to let the neural network learn several related tasks simultaneously (Caruan [7] These additional tasks are called catalytic hints (Abu Mostafa [1]) Theoretical analysis of some simple problems has shown that adding extra knowledge or hints to the learning process improves the performance of neural networks after and during learning. Abu Mostafa [1] and Barber and Saad [4] have shown that hints or extra knowledge improve the ....
....simultaneously (Caruan [7] These additional tasks are called catalytic hints (Abu Mostafa [1] Theoretical analysis of some simple problems has shown that adding extra knowledge or hints to the learning process improves the performance of neural networks after and during learning. Abu Mostafa [1] and Barber and Saad [4] have shown that hints or extra knowledge improve the generalization ability of neural networks. The hints we mentioned above are not very general. A more general approach to using additional information besides data pairs for network learning would be to express this ....
[Article contains additional citation context not shown here]
Abu-Mostafa, Y. S.: Hints and the VC dimension, Neural Computation 5(2) (1993a), 278--288.
On the Learnability of Rich Function Classes - Ratsaby, Maiorov (1983) (7 citations) (Correct)
....Example 1 (Parametric classification) The setting consists of two pattern classes with unknown class conditional probability densities f 1 (x) f 2 (x) over X=R and a priori probabilities p 1 , p 2 . Thedata sample consists of labeled examples [ x i , y i ) i=1 , where y i is first drawn from [1, 2], taking the value 1 with probability p 1 , and x i is drawn with respect to f y (x i ) 1#i#m.Thetarget g(x)is the discriminate function corresponding to the Bayes optimal classifier which classifies the region R g = x # X : g(x) ln( p 1 f 1 (x)#p 2 f 2 (x) #0] by 1 and the region X R g by 2. ....
Y. S. Abu-Mostafa, Hints and the VC dimension, Neural Comput. 5 (1993), 278#288.
Hints - Abu-Mostafa (1995) Self-citation (Abu-mostafa) (Correct)
....hypothesis. This is reflected in a bigger value of V C(G) When a hint is introduced, the VC dimension is affected. Since the hint is a valid property of f , we can use it as a litmus test to weed out bad g s thus shrinking G without losing good hypotheses. This leads to two new VC dimensions [3]: 1. The VC dimension provides an estimate for the number of examples needed to learn f , and since a hint H reduces the number of examples needed, a smaller VC dimension given the hint , V C(GjH) emerges. 2. If H itself is represented to the learning process by virtual examples, we can ask ....
Y. Abu-Mostafa, "Hints and the VC dimension," Neural Computation 5, pp. 278288, 1993.
Bayesian Applications of Belief Networks and.. - Antal, Fannes.. (2003) (Correct)
No context found.
Abu-Mostafa YS. Hints and the VC dimension. Neural Comput 1993;5(2):278--88.
Learning with Side Information: Part I - Kuusela, Ocone (2002) (Correct)
No context found.
Y. Abu-Mostafa, "Hints and the VC dimension," Neural Computation, vol. 5, pp. 278--288, 1993.
On the Value of Partial Information for Learning from Examples - Ratsaby, Maiorov (1998) (1 citation) (Correct)
No context found.
Abu-Mostafa, Y. S. (1993), Hints and the VC dimension, Neural Comput. 5, 278--288.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC