Incorporating a priori Knowledge into Initialized Weights for Neural Classifier
نویسندگان
چکیده
Artificial neural networks (ANN), esp. multilayer perceptrons (MLP) have been widely used in pattern recognition and classification. Nevertheless, how to incorporate a priori knowledge to design ANN is still an open problem. This paper tries to give some insightful discussions on this topic emphasizing weight initialization from three perspectives. Theoretical analyses and simulations are offered for validation.
منابع مشابه
Feasibility of Training and Investigation into Training of Sigmoidal FFANN with Gaussian Learning Rate and Zero Weight Initializations
Artificial Neural Networks(ANN) has applications in the various fields. Non linear transformation problems can be solved using ANN. Generally the parameter weight is initialized to some random value. In this paper we have used Sigmoidal FFANN. The training is done using Gaussian learning rate and weights are initialized to zero. It is found that the training of Sigmoidal FFANN can be done even ...
متن کاملIncorporating prior knowledge of predictors into penalized classifiers with multiple penalty terms
MOTIVATION In the context of sample (e.g. tumor) classifications with microarray gene expression data, many methods have been proposed. However, almost all the methods ignore existing biological knowledge and treat all the genes equally a priori. On the other hand, because some genes have been identified by previous studies to have biological functions or to be involved in pathways related to t...
متن کاملMRF-MBNN: A Novel Neural Network Architecture for Image Processing
Contextual information and a priori knowledge play important roles in image segmentation based on neural networks. This paper proposed a method for including contextual information in a model-based neural network (MBNN) that has the advantage of combining a priori knowledge. This is achieved by including Markov random field (MRF) into the MBNN and this novel neural network is termed as MRF-MBNN...
متن کاملOptimizing Cepstral Features for Audio Classification
Cepstral features have been widely used in audio applications. Domain knowledge has played an important role in designing different types of cepstral features proposed in the literature. In this paper, we present a novel approach for learning optimized cepstral features directly from audio data to better discriminate between different categories of signals in classification tasks. We employ mul...
متن کاملSmarter initializations in multi-modal neural networks to predict transcription factor binding
Various deep neural methods have been developed to predict transcription factor (TF) binding on DNA regions. In this paper we employ neural-network initialization techniques inspired by domain knowledge to augment current work in TF binding prediction. We train a Convolutional Neural Network (CNN) with the first layer of filters initialized in varying proportions to the values of position weigh...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2000