Srimanta Pal

SOFM-MLP: a hybrid neural network for atmospheric temperature prediction

Here, first we study the effectiveness of multilayer perceptron networks (MLPs) for prediction of the maximum and the minimum temperatures based on past observations on various atmospheric parameters. To capture the seasonality of atmospheric data, with a view to improving the prediction accuracy, we then propose a novel neural architecture that combines a self-organizing feature map (SOFM) and MLPs to realize a hybrid network named SOFM-MLP with better performance. We also demonstrate that the use of appropriate features such as temperature gradient can not only reduce the number of features drastically, but also can improve the prediction accuracy. These observations inspired us to use a feature selection MLP (FSMLP) instead of MLP, which can select good features online while learning the prediction task. FSMLP is used as a preprocessor to select good features. The combined use of FSMLP and SOFM-MLP results in a network system that uses only very few inputs but can produce good prediction.

A multi-stage neural network aided system for detection of microcalcifications in digitized mammograms

We propose a multi-stage detection system for microcalcification. A connectionist online feature selection technique is used to identify a set of good features from a set of 87 features computed at a few randomly selected positive (calcified) and negative (normal) pixels. A neural network is then trained with the selected features. The network output is cleaned using connected component analysis and an algorithm for removing thin elongated structures. A measure of local density (called mountain potential) of the calcified points is then computed at every suspected pixel of these cleaned images and the peak of the mountain potential is used to classify mammograms as calcified or normal. The system is tested on a set of 17 mammograms comprising 10 abnormal and seven normal images which are not used in training and the system is found to perform very well. Moreover for each abnormal image, the system is able to locate the calcified regions quite accurately.

Novel approach to identify good tracer clouds from a sequence of satellite images

A novel hierarchical method for finding tracer clouds from weather satellite images is proposed. From the sequence of cloud images, different features such as mean, standard deviation, busyness, and entropy are extracted. Based on these features, clouds are segmented using the k-means clustering algorithm and considering the coldest cloud segment, potential regions for tracer clouds are identified. These regions are represented by a set of features. All such steps are repeated for images taken at three consecutive time instants. Then, simulated annealing is used to establish an association between cloud segments of successive image frames. In this way, several chains of associated cloud regions are found and are ranked using fuzzy reasoning. The method has been tested in several image sequences, and its results are validated by determining cloud motion vector from the associated chains of tracers.

A soft computing approach for rainfall retrieval from the TRMM microwave imager

A neural network model for rainfall retrieval over ocean from remotely sensed microwave (MW) brightness temperature (BT) is proposed. BT data are obtained from the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI). The BT values from different channels of TMI over the Pacific Ocean (163/spl deg/ to 177/spl deg/W and 18/spl deg/ to 34/spl deg/S) are the input features. The near-surface rainfall rate from the Precipitation Radar (PR) are considered as a target. The proposed model consists of a neural network with online feature selection (FS) and clustering techniques. A K-means clustering algorithm is applied to cluster the selected features. Different networks have been trained to give an instantaneous rainfall rate with all input features as well as with selected features obtained by applying the FS algorithm. It is found that the hybrid network utilizing FS and clustering techniques performs better. The developed network is also validated with two independent datasets on March 14, 2000 over the Atlantic Ocean having stratiform rain and on March 21, 2000 over the Pacific Ocean having both stratiform and convective rain. In both cases, the hybrid network performs well with correlation coefficient improving to 0.78 and 0.81, respectively, in contrast to 0.70 and 0.75 for the network with all features. The rainfall rate retrieved from the hybrid network is also compared with the TMI surface rain rate, and a correlation of 0.84 and 0.75 is found for the two events. The proposed hybrid model is validated with a Doppler Weather Radar, and correlation of 0.52 is observed.

An Artificial-Neural-Network-Based Integrated Regional Model for Rain Retrieval Over Land and Ocean

An integrated regional model is proposed for rain-rate retrievals over land/ocean from the brightness temperature (Tb) values of the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI). The polarization-corrected temperature calculated from the 85.5-GHz channels is also considered as one of the inputs along with the nine channel Tb values. This model is applicable over the region between and . For this purpose, an artificial neural network is utilized. The collocated precipitation radar (PR) near-surface rain rates as given by a 2A25 data product is considered as a target value. The methodology consists of the separation of land and ocean pixels, the separation of stratiform and convective pixels over land/ocean, and the selection of important features (inputs) for the multilayer perceptron network by the feature selection technique for each group. For the separation of land/ocean pixels, the Tb values of the 10.65-GHz vertical channel are utilized. The values are utilized to separate the stratiform and convective pixels both over land and ocean. The rain retrieval from the developed model is validated with TRMM PR. Overall result shows the better agreement of the model-retrieved rain rate with the PR observation compared to the TMI (2A12) rain rate particularly over land. The rain retrieved from the developed model is further validated with Doppler weather radar. A reasonably good agreement is observed between these two estimations.

A Fusion Neural Network for Estimation of Blasting Vibration

This paper presents the effectiveness of multilayer perceptron (MLP) networks for estimation of blasting vibration using past observations of various blasting parameters. Here we propose a fusion network that combines several MLPs and on-line feature selection technique to obtain more reliable and accurate estimation over the empirical models.

A multilayer self-organizing model for convex-hull computation

A self-organizing neural-network model is proposed for computation of the convex-hull of a given set of planar points. The network evolves in such a manner that it adapts itself to the hull-vertices of the convex-hull. The proposed network consists of three layers of processors. The bottom layer computes some angles which are passed to the middle layer. The middle layer is used for computation of the minimum angle (winner selection). These information are passed to the topmost layer as well as fed back to the bottom layer. The network in the topmost layer self-organizes by labeling the hull-processors in an orderly fashion so that the final convex-hull is obtained from the topmost layer. Time complexity of the proposed model is analyzed and is compared with existing models of similar nature.

Inspection of 6X19 seale preformed haulage rope by nondestructive technique

Steel winding ropes used in aerial ropeways, deteriorate with use. Visual inspection is the most conventional method for evaluation of aerial ropes. But this method cannot evaluate the inner failure such as corrosion and other flaws. Nondestructive testing method is the only means for assessment of haulage ropes in terms of local faults (LF) and loss in metallic area (LMA). An attempt has been made in this paper to evaluate and monitor the condition of a 6X19 Seale preformed haulage rope by nondestructive technique.

Fuzzy Rule–Based Approach for Detection of Bounded Weak-Echo Regions in Radar Images

A method for the detection of a bounded weak-echo region (BWER) within a storm structure that can help in the prediction of severe weather phenomena is presented. A fuzzy rule–based approach that takes care of the various uncertainties associated with a radar image containing a BWER has been adopted. The proposed technique automatically finds some interpretable (fuzzy) rules for classification of radar data related to BWER. The radar images are preprocessed to find subregions (or segments) that are suspected candidates for BWERs. Each such segment is classified into one of three possible cases: strong BWER, marginal BWER, or no BWER. In this regard, spatial properties of the data are being explored. The method has been tested on a large volume of data that are different from the training set, and the performance is found to be very satisfactory. It is also demonstrated that an interpretation of the linguistic rules extracted by the system described herein can provide important characteristics about the underlying process.

A hybrid neural architecture and its application to temperature prediction

We first investigate the effectiveness of multilayer perceptron networks for prediction of atmospheric temperature. To capture the seasonality of atmospheric data we then propose a hybrid network, SOFM-MLP, that combines a self-organizing feature map (SOFM) and multilayer perceptron networks (MLPs). The architecture is quite general in nature and can be applied in other application areas. We also demonstrate that use of appropriate features can not only reduce the number of features but also can improve the prediction accuracies.