Abstract: Dunkl operator here we introduce a modified version of and use it to prove a theorem shows that functionals and rth order modulus of smoothness in K-theorem shows thatare equivalent. We use this equivalence to introduce p<1 spaces for Lp (K) essential degree of approximation using regular neural networks p and how a multivariate function in spaces for can be approximated using a p<1 spaces for Lp (K) multivariate p function in forward regular neural network. So, we can have the essential approximation using regular FFN. P<1 spaces for Lp (K) ability of a multivariate function in spaces for using regular FFN.