Conference Papers
Permanent URI for this collectionhttps://idr.nitk.ac.in/handle/123456789/28506
Browse
61 results
Search Results
Item Raga classification for Carnatic music(Springer Verlag service@springer.de, 2015) Suma, S.M.; Koolagudi, S.G.In this work, an effort has been made to identify raga of given piece of Carnatic music. In the proposed method, direct raga classification without the use of note sequence has been performed using pitch as the primary feature. The primitive features that are extracted from the probability density function (pdf) of the pitch contour are used for classification. A feature vector of 36 dimension is obtained by extracting some parameters from the pdf. Since non-sequential features are extracted from the signal, artificial neural network (ANN) is used as a classifier. The database used for validating the system consists of 162 songs from 12 ragas. The average classification accuracy is found to be 89.5%. © Springer India 2015.Item Distributed-Intrusion Detection System using combination of Ant Colony Optimization (ACO) and support vector machine (SVM)(Institute of Electrical and Electronics Engineers Inc., 2016) Wankhade, A.; Chandrasekaran, K.Intrusion Detection System (IDS) are playing a very substantial role in protecting computer networks. Still conventional IDS finds itself limited when it comes to distribute intrusion detection. An intruder may conceal its origin of attack by moving from node to node in a network. In order to conquer these limitations, alerts are to be exchanged and correlated in distributed intrusion detection system (DIDS) in a cooperative manner. Because of diversity of network behavior and high growth in development of new types of attacks, intrusion detection algorithm based on fast machine learning methods are of great significance to reduce the false alarm rates with high accuracy of detection rate. This work proposes using a DIDS model for data collection across the network and a hybrid method that classifies the network activities collected in the DIDS model as normal and abnormal. This hybrid method is a combination of popular machine learning algorithms Support Vector Machine (SVM) and Ant Colony Optimization (ACO) which is to be used on a model for DIDS. Also it can detect unseen attacks of intrusion with high detection rate with minimal misclassification. Experiments show that usage of hybrid method on the DIDS model is superior to that of SVM alone or ACO alone both in terms of run-Time efficiency and detection rate. © 2016 IEEE.Item Efficient audio segmentation in soccer videos(Institute of Electrical and Electronics Engineers Inc., 2016) Raghuram, M.A.; Chavan, N.R.; Koolagudi, S.G.; Ramteke, P.B.Identifying different audio segments in videos is the first step for many important tasks such as event detection and speech transcription. Approaches using Mel-Frequency Cepstral coefficients (MFCCs) with Gaussian mixture models (GMMs) and hidden Markov models (HMMs) perform reasonably well in stationary conditions but do not scale to a broad range of environmental conditions. This paper focuses on the audio segmentation in broadcast soccer videos into audio classes such as Silence, Speech Only, Speech Over Crowd, Crowd Only and Excited, with an alternative feature set which is simplistic as well as robust to changes in the environment conditions. Support Vector Machines (SVMs), Neural Networks and Random Forest are used for the classification. The accuracy achieved with SVMs, Neural Networks and Random Forest are 83.80%, 86.07%, and 88.35% respectively. The proposed features and Random Forest classifier are found to achieve better accuracy compared to the other classifiers. © 2016 IEEE.Item Image processing approach to diagnose eye diseases(Springer Verlag service@springer.de, 2017) Prashasthi, P.; Shravya, K.S.; Deepak, A.; Mulimani, M.; Shashidhar, K.G.Image processing and machine learning techniques are used for automatic detection of abnormalities in eye. The proposed methodology requires a clear photograph of eye (not necessarily a fundoscopic image) from which the chromatic and spatial property of the sclera and iris is extracted. These features are used in the diagnosis of various diseases considered. The changes in the colour of iris is a symptom for corneal infections and cataract, the spatial distribution of different colours distinguishes diseases like subconjunctival haemorrhage and conjunctivitis, and the spatial arrangement of iris and sclera is an indicator of palsy. We used various classifiers of which adaboost classifier which was found to give a substantially high accuracy i.e., about 95% accuracy when compared to others (k-NN and naive-Bayes). To enumerate the accuracy of the method proposed, we used 150 samples in which 23% were used for testing and 77% were used for training. © Springer International Publishing AG 2017.Item Temporal topic modeling of scholarly publications for future trend forecasting(Springer Verlag service@springer.de, 2017) Bhopale, A.P.; Kamath S․, S.S.The volume of scholarly articles published every year has grown exponentially over the years. With these growths in both core and interdisciplinary areas of research, analyzing interesting research trends can be helpful for new researchers and organizations geared towards collaborative work. Existing approaches used unsupervised learning methods such as clustering to group articles with similar characteristics for topic discovery, with low accuracy. Efficient and fast topic discovery models and future trend forecasters can be helpful in building intelligent applications like recommender systems for scholarly articles. In this paper, a novel approach to automatically discover topics (latent factors) from a large set of text documents using association rule mining on frequent itemsets is proposed. Temporal correlation analysis is used for finding the correlation between a set of topics, for improved prediction. To predict the popularity of a topic in the near future, time series analysis based on a set of topic vectors is performed. For experimental validation of the proposed approach, a dataset composed of 17 years worth of computer science scholarly articles, published through standard IEEE conferences was used, and the proposed approach achieved meaningful results. © Springer International Publishing AG 2017.Item What makes a video memorable?(Institute of Electrical and Electronics Engineers Inc., 2017) Kar, A.; Prashasthi, P.; Ghaturle, Y.; Vani, M.Humans are exposed to many pictures and videos on a daily basis, but they have this exceptional ability to remember the details, even though many of them look very similar. This Video Memorability (VM) is mainly due to distinguishable and a fine representation of the frames in human mind that people tend to remember. Videos have an abundance data contained in the frames which can be used for feature extraction purposes. Each feature from each frame has to be carefully considered to determine the intrinsic property of the video i.e. memorability. Using Convolutional Neural Network (CNN), we propose a solution to the problem of predicting VM, by estimating its memorability. A model has been developed to predict VM using algorithmically extracted features. Two types of features (i) semantic features (ii) visual features have been considered. The effectiveness of the model has been tested using publicly available image and video data. The results confirm that the CNN model can predict memorability with a acceptable performance. © 2017 IEEE.Item Protein secondary structural class prediction using effective feature modeling and machine learning techniques(Institute of Electrical and Electronics Engineers Inc., 2018) Bankapur, S.; Patil, N.Protein Secondary Structural Class (PSSC) prediction is an important step to find its further folds, tertiary structure and functions, which in turn have potential applications in drug discovery. Various computational methods have been developed to predict the PSSC, however, predicting PSSC on the basis of protein sequences is still a challenging task. In this study, we propose an effective approach to extract features using two techniques (i) SkipXGram bi-gram: in which skipped bi-gram features are extracted and (ii) Character embedded features: in which features are extracted using word embedding approach. The combined feature sets from the proposed feature modeling approach are explored using various machine learning classifiers. The best performing classifier (i.e. Random Forest) is benchmarked against state-of-the-art PSSC prediction models. The proposed model was assessed on two low sequence similarity benchmark datasets i.e. 25PDB and FC699. The performance analysis demonstrates that the proposed model consistently outperformed state-of-the-art models by a factor of 3% to 23% and 4% to 6% for 25PDB and FC699 datasets respectively. © 2018 IEEE.Item A Supervised Approach for Patient-Specific ICU Mortality Prediction Using Feature Modeling(Springer Verlag service@springer.de, 2019) S. Krishnan, G.S.; Kamath S․, S.K.Intensive Care Units (ICUs) are one of the most essential, but expensive healthcare services provided in hospitals. Modern monitoring machines in critical care units continuously generate huge amount of data, which can be used for intelligent decision-making. Prediction of mortality risk of patients is one such predictive analytics application, which can assist hospitals and healthcare personnel in making informed decisions. Traditional scoring systems currently in use are parametric scoring methods which often suffer from low accuracy. In this paper, an empirical study on the effect of feature selection on the feature set of traditional scoring methods for modeling an optimal feature set to represent each patient’s profile along with a supervised learning approach for ICU mortality prediction have been presented. Experimental evaluation of the proposed approach in comparison to standard severity scores like SAPS-II, SOFA and OASIS showed that the proposed model outperformed them by a margin of 12–16% in terms of prediction accuracy. © 2019, Springer Nature Singapore Pte Ltd.Item Predicting stock prices using ensemble learning and sentiment analysis(Institute of Electrical and Electronics Engineers Inc., 2019) Pasupulety, U.; Abdullah Anees, A.; Anmol, S.; Mohan, B.R.The recent success of the application of Artificial Intelligence in the financial sector has resulted in more firms relying on stochastic models for predicting the behaviour of the market. Everyday, quantitative analysts strive to attain better accuracies from their machine learning models for forecasting returns from stocks. Support Vector Machine (SVM) and Random Forest based regression models are known for their effectiveness in accurately predicting closing prices. In this work, we propose a technique for analyzing and predicting stock prices of companies using the aforementioned algorithms as an ensemble. Datasets from India's National Stock Exchange (NSE) containing basic market price information are preprocessed to include well known leading technical indicators as features. Feature selection, which ranks features based on their degree of influence on the final closing price has been incorporated to reduce the size of the training dataset. Additionally, we evaluate the effectiveness of considering the public opinion of a company by employing sentiment analysis. Using a trained Word2Vec model, company specific hash-tagged posts from Twitter are classified as positive or negative. Our proposed ensemble model is then trained on a new dataset which combines the technical indicator data along with the aggregated number of positive/negative tweets of a company over time. Our experiments indicate that in some scenarios, the ensemble model performs better than the constituent models and is highly dependent of the nature and size of the training data. However, combining technical indicator data with aggregated positive/negative tweet counts has a negligible effect on the performance of the ensemble model. © 2019 IEEE.Item Hybrid text feature modeling for disease group prediction using unstructured physician notes(Springer Science and Business Media Deutschland GmbH, 2020) S. Krishnan, G.S.; Kamath S․, S.Existing Clinical Decision Support Systems (CDSSs) largely depend on the availability of structured patient data and Electronic Health Records (EHRs) to aid caregivers. However, in case of hospitals in developing countries, structured patient data formats are not widely adopted, where medical professionals still rely on clinical notes in the form of unstructured text. Such unstructured clinical notes recorded by medical personnel can also be a potential source of rich patient-specific information which can be leveraged to build CDSSs, even for hospitals in developing countries. If such unstructured clinical text can be used, the manual and time-consuming process of EHR generation will no longer be required, with huge person-hours and cost savings. In this article, we propose a generic ICD9 disease group prediction CDSS built on unstructured physician notes modeled using hybrid word embeddings. These word embeddings are used to train a deep neural network for effectively predicting ICD9 disease groups. Experimental evaluation showed that the proposed approach outperformed the state-of-the-art disease group prediction model built on structured EHRs by 15% in terms of AUROC and 40% in terms of AUPRC, thus proving our hypothesis and eliminating dependency on availability of structured patient data. © Springer Nature Switzerland AG 2020.