Digital Commons Network^™

Deepcon-Pre: Improved Protein Contact Map Prediction Using Inverse Covariance And Deep Residual Networks, Nachammai Palaniappan

Theses

As with most domains where machine learning methods are applied, correct feature engineering is critical when developing deep learning algorithms for solving the protein folding problem. Unlike the domains such as computer vision and natural language processing, feature engineering is not rigorously studied towards solving the protein folding problem. A recent research has highlighted that input features known as precision matrix are most informative for predicting inter-residue contact map, the key for building three-dimensional models. In this work, we study the significance of the precision matrix feature when very deep residual networks are trained. Using a standard dataset of 3456 …

Protein Inter-Residue Distance Prediction Using Residual And Capsule Networks, Andrew Dillon

Theses

The protein folding problem, also known as protein structure prediction, is the task of building three-dimensional protein models given their one-dimensional amino acid sequence. New methods that have been successfully used in the most recent CASP challenge have demonstrated that predicting a protein's inter-residue distances is key to solving this problem. Various deep learning algorithms including fully convolutional neural networks and residual networks have been developed to solve the distance prediction problem. In this work, we develop a hybrid method based on residual networks and capsule networks. We demonstrate that our method can predict distances more accurately than the algorithms …

A Study Of Machine Learning And Deep Learning Models For Solving Medical Imaging Problems, Fadi G. Farhat

Theses

Application of machine learning and deep learning methods on medical imaging aims to create systems that can help in the diagnosis of disease and the automation of analyzing medical images in order to facilitate treatment planning. Deep learning methods do well in image recognition, but medical images present unique challenges. The lack of large amounts of data, the image size, and the high class-imbalance in most datasets, makes training a machine learning model to recognize a particular pattern that is typically present only in case images a formidable task.

Experiments are conducted to classify breast cancer images as healthy or …

A Comparative Study Of Russian Trolls Using Several Machine Learning Models On Twitter Data, Kannan Neten Dharan Kannan Neten Dharan

Theses

Ever since Russian trolls have been brought into light, their interference in the 2016 US Presidential elections has been monitored and studied thoroughly. These Russian trolls have fake accounts registered on several major social media sites to influence public opinions. Our work involves trying to discover patterns in these tweets and classifying them by using different machine learning approaches such as Support Vector Machines, Word2vec and neural network models, and then creating a benchmark to compare all the different models. Two machine learning models are developed for this purpose. The first one is used to classify any given specific tweet …

Deep Morphological Neural Networks, Yucong Shen

Theses

Mathematical morphology is a theory and technique applied to collect features like geometric and topological structures in digital images. Determining suitable morphological operations and structuring elements for a give purpose is a cumbersome and time-consuming task. In this paper, morphological neural networks are proposed to address this problem. Serving as a non-linear feature extracting layers in deep learning frameworks, the efficiency of the proposed morphological layer is confirmed analytically and empirically. With a known target, a single-filter morphological layer learns the structuring element correctly, and an adaptive layer can automatically select appropriate morphological operations. For high level applications, the proposed …

Using Long Short-Term Memory (Lstm) Recurrent Neural Network (Rnn) To Classify Network Attacks, Pramita Sree Muhuri

Theses

Cyber-attacks have increased greatly in recent years. Therefore, the identification of various network attacks has been an important research area. An Intrusion Detection System (IDS), can identify an ongoing invasion or an intrusion which has already occurred. Intrusion Detection is a classification problem. It identifies whether the network traffic behavior is normal or anomalous or identifies the attack types. Various approaches have been proposed to improve the accuracy of classifiers for identifying the intrusion types. Recently, deep learning has emerged as a successful approach in IDSs having a high accuracy rate with its distinctive learning mechanism. In this research, Long …

Collaboration Pattern Model For Student Participation In Problem-Solving Typed Chat, Duy Quang Bui

Theses

This project measures different collaborative dialogue acts between students who are working together to solve problems in a computer programming class. In COMPS (Computer-Mediated Problem Solving) exercises students work together via online typed-chat. Transcripts of these conversations were annotated with four categories of collaborative utterance: sharing ideas, negotiating ideas, regulating problem-solving, and maintaining communication. The annotated transcripts were then applied to answer four different research questions. A) Among the several students in a conversation, there are measurable quantitative differences in dialogue behavior that correlate with the relative preparedness for solving the problem. The most prepared student not only talks more …

Music Retrieval System Using Dynamic Time Warping, Emeka Jude Okafor

Theses

With the growth of digital audio data, various and fast access to music data is strongly desired, especially for large music databases. A more natural way to retrieve a song from a database will be to hum to the tune. To relate and compare musical pieces is a very complex task. Musical compositions usually collapse multiple information sources and complex, multifaceted interactions established between parts. Despite such degrees of complexity, humans are outstandingly good at performing individual musical judgments with little conscious effort, while a computer cannot efficiently achieve this task. In this work, we focus on one such task: …