“…As to the TaylorNet, it is also worth mentioning that our proposal is very different from the previous work in the literature [91][92][93][94][95], which have also explored the Taylor series in neural networks. Most of them have specific motivations and work in different frameworks, and there is no explicit hierarchical structure employed there.…”
We propose to employ the hierarchical coarse-grained structure in the artificial neural networks explicitly to improve the interpretability without degrading performance. The idea has been applied in two situations. One is a neural network called TaylorNet, which aims to approximate the general mapping from input data to output result in terms of Taylor series directly, without resorting to any magic nonlinear activations. The other is a new setup for data distillation, which can perform multilevel abstraction of the input dataset and generate new data that possesses the relevant features of the original dataset and can be used as references for classification. In both cases, the coarse-grained structure plays an important role in simplifying the network and improving both the interpretability and efficiency. The validity has been demonstrated on MNIST and CIFAR-10 datasets. Further improvement and some open questions related are also discussed.
“…As to the TaylorNet, it is also worth mentioning that our proposal is very different from the previous work in the literature [91][92][93][94][95], which have also explored the Taylor series in neural networks. Most of them have specific motivations and work in different frameworks, and there is no explicit hierarchical structure employed there.…”
We propose to employ the hierarchical coarse-grained structure in the artificial neural networks explicitly to improve the interpretability without degrading performance. The idea has been applied in two situations. One is a neural network called TaylorNet, which aims to approximate the general mapping from input data to output result in terms of Taylor series directly, without resorting to any magic nonlinear activations. The other is a new setup for data distillation, which can perform multilevel abstraction of the input dataset and generate new data that possesses the relevant features of the original dataset and can be used as references for classification. In both cases, the coarse-grained structure plays an important role in simplifying the network and improving both the interpretability and efficiency. The validity has been demonstrated on MNIST and CIFAR-10 datasets. Further improvement and some open questions related are also discussed.
“…Take the Taylor series of the unary function as an example. In order to describe the finite Taylor series, considering the sum of the first + 1 terms, we can get formula (5).…”
Section: Neural Network Modelmentioning
confidence: 99%
“…Explicit polynomial fitting often uses Taylor series [5,6]. Taylor series are often used in the field of mathematics, especially in the research of approximate calculations.…”
As a method of function approximation, polynomial fitting has always been the main research hotspot in mathematical modeling. In many disciplines such as computer, physics, biology, neural networks have been widely used, and most of the applications have been transformed into fitting problems using neural networks. One of the main reasons that neural networks can be widely used is that it has a certain sense of universal approximation. In order to fit the polynomial, this paper constructs a three-layer feedforward neural network, uses Taylor series as the activation function, and determines the number of hidden layer neurons according to the order of the polynomial and the dimensions of the input variables. For explicit polynomial fitting, this paper uses non-linear functions as the objective function, and compares the fitting effects under different orders of polynomials. For the fitting of implicit polynomial curves, the current popular polynomial fitting algorithms are compared and analyzed. Experiments have proved that the algorithm used in this paper is suitable for both explicit polynomial fitting and implicit polynomial fitting. The algorithm is relatively simple, practical, easy to calculate, and can efficiently achieve the fitting goal. At the same time, the computational complexity is relatively low, which has certain application value.
“…As to the TaylorNet, it is also worth mentioning that our proposal is very different from the previous work in the literature, [91][92][93][94][95][96][97] which have also explored the Taylor series in neural networks. Most of them have specific motivations and work in different frameworks, and there is no explicit hierarchical structure employed there.…”
mentioning
confidence: 96%
“…Most of them have specific motivations and work in different frameworks, and there is no explicit hierarchical structure employed there. For example, Chen et al [91] used a single-layer neural network similar to Eq. ( 3) to approximate the Taylor expansion of a single-variable function.…”
We propose to employ the hierarchical coarse-grained structure in the artificial neural networks explicitly to improve the interpretability without degrading performance. The idea has been applied in two situations. One is a neural network called TaylorNet, which aims to approximate the general mapping from input data to output result in terms of Taylor series directly, without resorting to any magic nonlinear activations. The other is a new setup for data distillation, which can perform multi-level abstraction of the input dataset and generate new data that possesses the relevant features of the original dataset and can be used as references for classification. In both cases, the coarse-grained structure plays an important role in simplifying the network and improving both the interpretability and efficiency. The validity has been demonstrated on MNIST and CIFAR-10 datasets. Further improvement and some open questions related are also discussed.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.