Disjunctive normal networks
نویسندگان
چکیده
Artificial neural networks are powerful pattern classifiers. They form the basis of the highly successful and popular Convolutional Networks which offer the state-of-the-art performance on several computer visions tasks. However, in many general and non-vision tasks, neural networks are surpassed by methods such as support vector machines and random forests that are also easier to use and faster to train. One reason is that the backpropagation algorithm, which is used to train artificial neural networks, usually starts from a random weight initialization which complicates the optimization process leading to long training times and increases the risk of stopping in a poor local minima. Several initialization schemes and pre-training methods have been proposed to improve the efficiency and performance of training a neural network. However, this problem arises from the architecture of neural networks. We use the disjunctive normal form and approximate the boolean conjunction operations with products to construct a novel network architecture. The proposed model can be trained by minimizing an error function and it allows an effective and intuitive initialization which avoids poor local minima. We show that the proposed structure provides efficient coverage of the decision space which leads to state-of-the art classification accuracy and fast training times. & 2016 Elsevier B.V. All rights reserved.
منابع مشابه
Relational Learning of Disjunctive Patterns in Spatial Networks
In spatial domains, objects present high heterogeneity and are connected by several relationships to form complex networks. Mining spatial networks can provide information on both the objects and their interactions. In this work we propose a descriptive data mining approach to discover relational disjunctive patterns in spatial networks. Relational disjunctive patterns permit to represent spati...
متن کاملϵ-OA for the solution of bi-objective generalized disjunctive programming problems in the synthesis of nonlinear process networks
متن کامل
Tree Decomposition of Disjunctive Fuzzy Temporal Constraint Networks
The Disjunctive Fuzzy Temporal Constraint Networks model allows to express complex temporal constraints, but its general complexity is exponential. There is a subset of these networks, having a tree structure, which can be solved in polinomial time. Thus, it can be interesting to decide if a Disjunctive Fuzzy Temporal Constraint Network can be reduced to an equivalent network with tree structur...
متن کاملPolyhedrons and Perceptrons Are Functionally Equivalent
Mathematical definitions of polyhedrons and perceptron networks are discussed. The formalization of polyhedrons is done in a rather traditional way. For networks, previously proposed systems are developed. Perceptron networks in disjunctive normal form (DNF) and conjunctive normal forms (CNF) are introduced. The main theme is that single output perceptron neural networks and characteristic func...
متن کاملA Tractable Subclass of Fuzzy Constraint Networks
The Fuzzy Constraint Networks model, a generalization of the Disjunctive Temporal Fuzzy Constraint Networks, is a framework that allows representing and reasoning with fuzzy qualitative and quantitative complex constraints. However, its general complexity is exponential, and we need to find tractable subclasses. In this paper we propose two algorithms to deal with a tractable subclass named Ser...
متن کاملNonlinear disjunctive kriging for the estimating and modeling of a vein copper deposit
ABSTRACT Estimation of mineral resources and reserves with low values of error is essential in mineral exploration. The aim of this study is to estimate and model a vein type deposit using disjunctive kriging method. Disjunctive Kriging (DK) as an appropriate nonlinear estimation method has been used for estimation of Cu values. For estimation of Cu values and modelling of the distributio...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید
ثبت ناماگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید
ورودعنوان ژورنال:
- Neurocomputing
دوره 218 شماره
صفحات -
تاریخ انتشار 2016