publications

3D Segmentation, Slice Propagation, Uncertainty Quantification

Supervised methods for 3D anatomy segmentation demonstrate superior performance but are often limited by the availability of annotated data. This limitation has led to a growing interest in self-supervised approaches in tandem with the abundance of available un-annotated data. Slice propagation has emerged as an self-supervised approach that leverages slice registration as a self-supervised task to achieve full anatomy segmentation with minimal supervision. This approach significantly reduces the need for domain expertise, time, and the cost associated with building fully annotated datasets required for training segmentation networks. However, this shift toward reduced supervision via deterministic networks raises concerns about the trustworthiness and reliability of predictions, especially when compared with more accurate supervised approaches. To address this concern, we propose the integration of calibrated uncertainty quantification (UQ) into slice propagation methods, providing insights into the model's predictive reliability and confidence levels. Incorporating uncertainty measures enhances user confidence in self-supervised approaches, thereby improving their practical applicability. We conducted experiments on three datasets for 3D abdominal segmentation using five UQ methods. The results illustrate that incorporating UQ improves not only model trustworthiness, but also segmentation accuracy. Furthermore, our analysis reveals various failure modes of slice propagation methods that might not be immediately apparent to end-users. This study opens up new research avenues to improve the accuracy and trustworthiness of slice propagation methods.

Machine Learning, Prediction Model, Movie Success Prediction, Support Vector Machine, Naive Bayes, Logistic Regression

The entertainment industry is a rapidly growing billion-dollar industry. With new milestones being reached almost every day, this industry has proved itself to be a very profitable business, if done correctly. Since huge investments are involved in the production and making of movies, both in terms of time and money, it would only make sense to try to predict the outcome beforehand. In an attempt to tackle this problem, we have built a model that predicts whether or not a movie can be called a success. The model compares the performance of three machine learning algorithms i.e. Naive Bayes, Logistic Regression, and Support Vector Machine (SVM), over two different datasets, to observe which performs better. We have illustrated the model, as well as its results, findings, and observations in this literature.

Machine Learning, Deep Learning, Autoencoders, Face Mask, Unmasking

Machine Learning, Music Genre Classification, Decision Trees, K Nearest Neighbours, Logistic regression, Naïve Bayes, Neural Networks, Random Forest, Support Vector Machine, XGBoost

As Plato once rightfully said, ‘Music gives a soul to the universe, wings to the mind, flight to the imagination and life to everything.’ Music has always been an important art form, and more so in today’s science-driven world. Music genre classification paves the way for other applications such as music recommender models. Several approaches could be used to classify music genres. In this literature, we aimed to build a machine learning model to classify the genre of an input audio file using 8 machine learning algorithms and determine which algorithm is the best suitable for genre classification. We have obtained an accuracy of 91% using the XGBoost algorithm.

Sign Language, Machine Learning, Interpretation model, Convoluted Neural Networks, American Sign Language

Sign Language is invaluable to hearing and speaking impaired people and is their only way of communicating among themselves. However, it has limitations with its reach as the rest of the people have no information regarding sign language interpretation. Sign language is communicated via hand gestures and visual modes and is therefore used by hearing and speaking impaired people to intercommunicate. These languages have alphabets and grammar of their own, which cannot be understood by people who have no knowledge about the specific symbols and rules. Thus, it has become essential for everyone to interpret, understand and communicate via sign language to overcome and alleviate the barriers of speech and communication. This can be tackled with the help of machine learning. This model is a Sign Language Interpreter that uses a dataset of images and interprets the sign language alphabets and sentences with 90.9% accuracy. For this paper, we have used an ASL (American Sign Language) Alphabet. We have used the CNN algorithm for this project. This paper ends with a summary of the model’s viability and its usefulness for interpretation of Sign Language.

Deep Learning, Convolutional Neural Network, Image Colorization, Autoencoders

Modification of art may be viewed as enhancement or vandalization. Even though for a long time many were opposed to the idea of colorizing images, they now have finally viewed it for what it is - an enhancement of the art form. Grayscale image colorization has since been a long-standing artistic division. It has been used to revive or modify images taken prior to the invention of colour photography. This paper explores one method to reinvigorate grayscale images by colorizing them. We propose the use of deep learning, specifically the use of convolution neural networks. The obtained results show the ability of our model to realistically colorize grayscale images.

Machine Learning, Binary Classification, Boosting Algorithm, Gradient boosting, XGBoost, AdaBoost, CatBoost

In the medical field, predicting a heart disease has become a very complicated and challenging task. So, in this contemporary lifestyle, there is an urgent need for a system that will help predict accurately the possibility of getting heart disease. This paper presents an observation-based comparison between four boosting algorithms namely Gradient boosting, XGBoost, ADAboost and CatBoost to predict heart failure efficiently. To do so, we have referred to the PLOS (Public Library of Science) Repository dataset. These algorithm’s performances have been evaluated using metrics like Accuracy, F1 score, Recall and many more. All values obtained ensured the superiority of these boosting algorithms based on several performance measures.

Machine Learning, Binary Classification, Prediction Model, Support Vector Machine, K Nearest Neighbours

Breast cancer is a greatly widespread and dangerous type of cancer with approximately 2.3 million cases in the past year. It has surpassed lung cancer which was the most common cancer. Data mining and classification of data have helped medical experts to segregate and make use of the data achieve a higher accuracy of results. In this paper, we have referred to the Wisconsin Breast Cancer dataset. We have compared SVM and kNN algorithms for training the dataset and the more accurate one is utilised to test the final model using 10-fold cross validation practice. Furthermore, the maximum accuracy was obtained by the SVM model in the training phase and was further tuned in order to achieve a final accuracy of 96.93% in the test phase.