教学课程资源库 更多>>

关键词： African Americans   Computer science   Machine learning   Osteoporosis   SNPs  

摘要： Osteoporosis is a debilitating disease in which an individual’s bones weaken, making bones fragile and more susceptible to fracture. While commonly found amongst postmenopausal Caucasian and Asian women based on previous studies, those of African descent (African American/Black) have largely been ignored when it comes to osteoporotic studies, especially when it comes to Genome Wide Association Studies (GWAS). From GWA studies, we gain access to single nucleotide poly-morphisms (SNPs) that may contribute to certain illnesses, such as osteoporosis. With low Bone Mineral Density (BMD) being one of the primary markers of potential osteoporosis, it is prudent that proper research is done in order to help the African American population circumvent or mitigate the worst symptoms and complications of osteoporosis. In this thesis, we implemented and applied machine learning algorithms to analyze genetic data of African American women in order to make predictions that map SNPs to BMD. Using Coefficient of Determination (R2) and Mean Squared Error (MSE) for evaluation, the machine learning techniques we utilized for this regression task are: regularized linear regression (Ridge, Lasso, and ElasticNet), gradient boosted trees (XGBoost and LightGBM), and artificial neural networks. With these models, we performed an analysis on 3 datasets, compromised of 12,600, 69,476, and 158,444 variants respectively. The first dataset, known as SNP-1, received its highest overall test R2 score of 0.227 through Lasso Regression. The second dataset, SNP-5, received its highest overall test R2 score of 0.437 through LightGBM. And lastly, the third dataset, SNP-10, received its highest overall test R2 score of 0.574 from Ridge Regression.

关键词： Computer science   Artificial intelligence   Information technology  

关键词： Aerospace engineering   Design   Engineering   Computer science  

摘要： The utility of Small Satellites (SmallSats) for technology demonstrations and scientific research has been proven over the past few decades by governments, universities, and private companies. While the research and technology demonstration objectives that can be provided by these SmallSats are becoming similar to larger spacecraft, their reliability still falls behind. This is in part due to the reduced cost of SmallSat missions in comparison to large spacecraft, which requires cheaper components, rapid development schedules, and accepted risk. In these missions, the importance of the flight software is often overlooked, and the software is rushed through development and not fully tested to provide the reliability required for on-orbit operations. This research aims to investigate the common causes of failures on SmallSats, and to provide a solution to the problem of developing and testing reliable flight software, through the use of software-based simulation of the full embedded satellite system. Through the course of this research, an open-source product was developed and released to the public to assist SmallSat missions, which is currently in use by public and private institutions across the country. The resulting product, the NASA Operational Simulator for Small Satellites, commonly referred to as NOS3, will be discussed in detail. The results of NOS3 will be viewed through a case study of the application of NOS3 to a SmallSat mission.

关键词： Computer science   Computer engineering   Applied mathematics   Optics  

关键词： Bioinformatics   Alternative Medicine   Bioengineering   Computer science   Genetics   Medicine   Microbiology   Morphology   Pharmaceutical sciences  

摘要： Natural products originating from microorganisms are frequently used in antimicrobial and anticancer drugs, pesticides, herbicides, or fungicides. In the last years, the increasing availability of microbial genome data has made it possible to access the wealth of biosynthetic clusters responsible for the production of these compounds by genome mining. antiSMASH is one of the most popular tools in this field. The antiSMASH database provides pre-computed antiSMASH results for many publicly available microbial genomes and allows for advanced cross-genome searches. The current version 2 of the antiSMASH database contains annotations for 6,200 full bacterial genomes and 18,576 bacterial draft genomes and is available at https://***/

关键词： Applied mathematics   Cellular biology   Artificial intelligence   Software   Datasets   Neural networks   Computer science  

关键词： Artificial intelligence   Computer science   Energy   Fashion models   Accuracy   Datasets   Graph representations   Optimization   Neural networks   Classification   Visualization  

摘要： Deep neural networks (DNNs) have achieved significant performance improvement in image classification tasks.A DNN utilizes an enormous number of trainable parameters and non-linear functions to learn a mapping from a set of input images to a set of target classes. However, the increasing number of parameters, called the overparameterization problem, can create major challenges in implementation of these models on resource limited devices and can encourage overfitting of the model. Pruning less important trainable parameters in DNNs is one of the major established methods to defeat the above challenges. Pruning methods are also applicable on small models by design to further reduce their size. This thesis proposes the concept of energy models for pruning convolutional filters and hidden units with corresponding incoming and outgoing connections and bias terms for image classification tasks. Specifically, two energy models, namely EPruning and IPruning, are proposed. The first approach uses the Hamiltonian of a given DNN to form an energy loss minimization problem. The loss function is defined based on the difference between the energy of target class and the closest energy of the competitive classes. The second approach uses entropy of feature maps, Kullback–Leibler (KL) divergence between convolution filters in a layer, and activation level of hidden units to form an Ising energy-based minimization problem for detecting less important and redundant parameters in the network. Despite most pruning methods, which are based on defining a pruning threshold per layer, the proposed methods are data-driven and perform a network-wise pruning of the trainable parameters. We define pruning as searching for a sub-network of a given DNN, which has the lowest energy loss value based on a defined energy loss function. Since this is an NP-hard combinatorial optimization problem, a population-based global optimization framework is proposed to minimize the pruning objective functions

关键词： Computer science   Artificial intelligence  

摘要： Mobile apps are at the center of everyone's daily lives and users give them access to their intimate personal data. Some apps collect more information than they need to perform their job. These apps are called intrusive, and can represent not only a privacy issue, but a security problem for their users. Therefore, it is important to develop methods for figuring out how much an app can detect and collect from its users, and whether that access is in line with their privacy expectations. Several methods have been devised to determine app intrusiveness, a measure that represents how much an app's data collection deviates from its basic needs. This number, called intrusiveness or privacy score, can guide a user in the process of identifying apps that gather too much personal information. Some of the methods to calculate intrusiveness include analysis of app descriptions and conformity with their programmed behavior. However, most of the existing approaches depend on static analysis that is quite challenging and mandates access to the binaries or source code. This thesis proposes a novel method to determine whether an app is intrusive based on its description, which can allow users to make decisions before *** specifically, we used a Long Short-Term Memory (LSTM) network to analyze the descriptions, along with a Multi-Layer Perceptron (MLP) network to process metadata provided by other app features. The results show that this combined network structure achieved 79% accuracy in training and 74% accuracy for validation, with 840,000 samples and a 75/25 split between training and validation. Our findings indicate that not only it is possible to use the description and other information available from the app store to predict the intrusiveness of an app, but also that the network required to do the job is fairly small.

关键词： Sample size   Behavior   Vortices   Sample variance   Aerodynamics   Automobiles   Algorithms   Feedback   High speed   Vehicles   Aerospace engineering   Computer science   Transportation  

摘要： The current automotive era is moving towards electrified vehicle propulsion. As a result, an energy efficient vehicle design becomes one of the top priorities. From an aerodynamics point of view, the vehicle should be more streamlined for minimal aerodynamic drag. Such designs have the potential to enhance vehicle sensitivity when exposed to external disturbances such as unsteady aerodynamic forces and moments created by the flow of air around the vehicle. Before signing off for production, several on-road test scenarios are conducted by professional drivers to evaluate the new vehicle’s performance. Finding vehicle instabilities and proposing solutions to problem’s during such late phases of development is challenging in many *** objective of this paper is to correlate and predict the driver’s subjective perception on high-speed straight-line driving stability with measurable quantities in the early phase of development. In this work, different aerodynamic devices were used for generating higher lift and asymmetric aerodynamic forces resulting in substandard straight-line drivability on-road. An inverted wing, an inverted wing with an asymmetric flat plate, and an asymmetric air curtain attached under the bumper were the selected aerodynamic devices paired with and without bumper side-kicks. The side-kicks help define the flow separation, thus improving the drivability of the tested vehicle. Plots of mean and standard deviation and ride diagrams of lateral acceleration, yaw velocity, steering angle, and steering torque are used to understand vehicle behaviour for the paired configurations and relate to the difference of subjective judgment of drivability within each pair. The ride diagram was used to separate the presence of transient behaviour and study its impact on subjective judgement. The qualitative assessment of the resulting trends agrees well with the subjective judgement of the driver. Clinical tests were conducted using driving simulators, in ord

关键词： Mathematics   Computer science   Applied mathematics  

摘要： We call an induced cycle of length at least four a hole. The parity of a hole is the parity of its *** holes of certain types in a graph has deep structural *** 2006, Chudnovksy, Seymour, Robertson, and Thomas famously proved that a graph is perfect if and only if it does not contain an odd hole or a complement of an odd *** 2002, Conforti, Cornuéjols, Kapoor and Vušković provided a structural description of the class of even-hole-free graphs. In Chapter 3, we provide a structural description of all graphs that contain only holes of length ℓ for every ℓ ≥ 4. Analysis of how holes interact with graph structure has yielded detection algorithms for holes of various lengths and *** 1991, Bienstock showed it is NP-Hard to test whether a graph G has an even (or odd) hole containing a specified vertex v∈V(G). In 2002, Conforti, Cornuéjols, Kapoor and Vušković gave a polynomial-time algorithm to recognize even-hole-free graphs using their structure theorem. In 2003, Chudnovsky, Kawarabayashi and Seymour provided a simpler and slightly faster algorithm to test whether a graph contains an even hole. In 2019, Chudnovsky, Scott, Seymour and Spirkl provided a polynomial-time algorithm to test whether a graph contains an odd hole. Later that year, Chudnovsky, Scott and Seymour strengthened this result by providing a polynomial-time algorithm to test whether a graph contains an odd hole of length at least ℓ for any fixed integer ℓ ≥ 5. In Chapter 2, we provide a polynomial-time algorithm to test whether a graph contains an even hole of length at least ℓ for any fixed integer ℓ ≥ 4.