No description available.

Machine learning is fast growing field as it can be applied to solve a large amount of problems. One large subsection of machine learning are artificial neural networks (ANN), these work on pattern recognition and can be trained with data sets of known solutions. The objective of this thesis is to discuss the creation of an ANN capable of classifying differences in propylene glycol concentrations, up to 10%. Utilizing a micro pipette thermal sensor (MTS) it is possible to measure the heat propagation of a liquid from a laser pulse. The ANN can then be trained beforehand with simulated data and be tested in real time with temperature data from the MTS. This method could be applied to find the thermal conductivity of unknown fluids and biological samples, such as cells and tissues.

No / Drug vehicles are chemical carriers that aid a drug's passage through an organism. Whilst they possess no intrinsic efficacy they are designed to achieve desirable characteristics which can include improving a drug's permeability and or solubility, targeting a drug to a specific site or reducing a drug's toxicity. All of which are ideally achieved without compromising the efficacy of the drug. Whilst the majority of drug vehicle research is focused on the solubility and permeability issues of a drug, significant progress has been made on using vehicles for toxicity reduction. Achieving this can enable safer and more effective use of a potent drug against diseases such as cancer. From a molecular perspective, drugs activate or deactivate biochemical pathways through interactions with cellular macromolecules resulting in toxicity. For newly developed drugs such pathways are not always clearly understood but toxicity endpoints are still required as part of a drug's registration. An understanding of which vehicles could be used to ameliorate the unwanted toxicities of newly developed drugs would be highly desirable to the pharmaceutical industry. In this paper we demonstrate the use of different classifiers as a means to select vehicles best suited to avert a drug's toxic effects when no other information about a drug's characteristics is known. Through analysis of data acquired from the Developmental Therapeutics Program (DTP) we are able to establish a link between a drug's toxicity and vehicle used. We demonstrate that classification and selection of the appropriate vehicle can be made based on the similarity of drug choice.

Interest in nanophotonics continues to grow as integrated optics provides an affordable platform for areas like telecommunications, quantum information processing, and biosensing. Designing and characterizing integrated photonics components and circuits, however, remains a major bottleneck. This is especially true when complex circuits or devices are required to study a particular phenomenon.To address this challenge, this work develops and experimentally validates a novel machine learning design framework for nanophotonic devices that is both practical and intuitive. As case studies, artificial neural networks are trained to model strip waveguides, integrated chirped Bragg gratings, and microring resonators using a small number of simple input and output parameters relevant to designers. Once trained, the models significantly decrease the computational cost relative to traditional design methodologies. To illustrate the power of the new design paradigm, both forward and inverse design tools enabled by the new design paradigm are demonstrated. These tools are directly used to design and fabricate several integrated Bragg grating devices and ring resonator filters. The method's predictions match the experimental measurements well and do not require any post-fabrication training adjustments.

Soil moisture is an important variable in many environmental researches and application areas as it affects the interactions between atmosphere and land surface by controlling the energy and water exchange. The current measurement techniques are insufficient to acquire accurate large-scale root zone soil moisture (RZSM) data at the spatial resolution of interest. Though assorted models have been successfully applied in relatively small areas to estimate RZSM, the large-scale estimation is still facing challenges as it requires the flexibility and practicality of the models for the applications under various conditions. Though physically based soil moisture models are widely used, the errors in model physics affect the flexibility of these models meanwhile their large demand of data and computational resources reduces the practicality. On the contrary, the statistical and data-driven methods have high potential but their applications for large-scale RZSM estimation have not been fully explored. To develop feasible models for large-scale RZSM estimation using the surface observations, artificial neural networks, specifically multilayer perceptrons (MLPs), were applied in this study to estimate RZSM at the depths of 20cm and 50cm, using the data of 557 stations in the United States. Two experiments including four models were developed and the input variables of the models were carefully selected. The sensitivity analysis found that surface soil moisture and the cumulative rainfall, snowfall, air temperature and surface soil temperature were important inputs. If given soil texture data as inputs, the models achieved better performance and were extremely sensitive to them. The results showed that the MLPs were effective and flexible for the estimation of soil moisture at 20cm under various climate types and were insensitive to the potential errors in soil moisture datasets. However, the results of the estimation at 50cm are not as good as that of the 20cm. / Thesis / Master of Science (MSc)

Искусственные нейронные сети – это концепция, которая исследуется с середины XX века, но до сих пор но только сейчас они переживают очень высокий темп роста. Благодаря значительным улучшениям в их поведения, за последние годы их использование перешло от использования только в академических целях до полностью внедрено и функционирует в нашей жизни. Эти нейронные сети являются системами, которые используются во многих различных приложениях в настоящее время. Таким образом, это дает нам главная особенность нейронных сетей: эти системы легко построить, самая большая проблема заключается в реализации алгоритма обучения, который состоит из следующих элементов алгоритм обучения, который состоит из нескольких очень простых итеративных математических операций (даже меньше, если мы используем и, в тоже время, это очень мощные системы. / Artificial neural networks are a concept that has been researched since the middle of the 20th century, but until now, but only now, they are experiencing a very high rate of growth. Due to significant improvements in their behavior, in recent years their use has gone from being used for academic purposes only to being fully implemented and functioning in our lives. These neural networks are systems that are used in many different applications nowadays. Thus, this gives us the main feature of neural networks: these systems are easy to build, the biggest problem is to implement a learning algorithm, which consists of the following elements, a learning algorithm that consists of several very simple iterative mathematical operations (even less if we use and At the same time, these are very powerful systems.

Fibre reinforcement polymers (FRP) are non-corrodible materials used instead of conventional steel and have been approved to be an effective way to overcome corrosion problems. FRP, in most cases, can have a higher tensile strength, but a lower tensile modulus of elasticity compared to that of conventional steel bars. This study aimed to examine flat slab specimens reinforced with glass fibre reinforced polymer (GFRP) and steel bar materials for punching shear behaviour. Six full-scale two-way slab specimens were constructed and tested under concentric load up to failure. One of the main objectives is to study the effect of reinforcement spacing with the same reinforcement ratio on the punching shear strength. In addition, two other parameters were considered, namely, slab depth, and compressive strength of concrete. The punching shear provisions of two code of practises CSA S806 (Canadian Standards 2012) and JSCE (JSCE et al. 1997) reasonably predicted the load capacity of GFRP reinforced concrete flat slab, whereas, ACI 440 (ACI Committee 440 2015) showed very conservative load capacity prediction. On the other hand, a dynamic explicit solver in nonlinear finite element (FE) modelling is used to analyse a connection of column to concrete flat slabs reinforced with GFRP bars in terms of ultimate punching load. All FE modelling was performed in 3D with the appropriate adoption of element size and mesh. The numerical and experimental results were compared in order to evaluate the developed FE, aiming to predict the behaviour of punching shear in the concrete flat slab. In addition, a parametric study was created to explore the behaviour of GFRP reinforced concrete flat slab with three parameters, namely, concrete strength, shear load perimeter to effective depth ratio, and, flexural reinforcement ratio. It was concluded that the developed models could accurately capture the behaviour of GFRP reinforced concrete flat slabs subjected to a concentrated load. Artificial Neural Networks (ANN) is used in this research to predict punching shear strength, and the results were shown to match more closely with the experimental results. A parametric study was performed to investigate the effects of five parameters on punching shear capacity of GFRP reinforced concrete flat slab. The parametric investigation revealed that the effective depth has the most substantial impact on the load carrying capacity of the punching shear followed by reinforcement ratio, column perimeter, the compressive strength of the concrete, and, the elastic modulus of the reinforcement.

This research investigates software failure and fault analysis through data-driven machine learning approaches. Faults can happen in any software system and may hugely impact system reliability and user experience. Log data, the machine-generated data that records the system status, is often the primary source of information to track down a fault. This study aims to develop automated systems that recognize recurring faults by analyzing the system log data. The methodology developed in this research applies to the Ford SYNC vehicle infotainment system as well as other systems that produce similar log data. Log data has been used in manual examination to help trace and localize a fault. This manual process can be effective and sometimes the only feasible way of troubleshooting software faults. However, as the amount of log data increases significantly with the growing complexity and scale of software, the manual workload can get overwhelming. During the system-level validation tests, all system components are producing log data, resulting in tens of thousands of lines of log messages in just a few minutes. Therefore, automated diagnosis has been a promising approach for log data analysis. Three machine learning approaches are investigated in this research to tackle the fault diagnosis problem: 1) the data mining approach; 2) the statistical feature approach; and, 3) the deep learning approach. The first method attempts to mimic human experts to examine log data. Log sequences representing a fault are extracted through data mining techniques and used to identify anomalies. The method is effective when applied to a small volume of data, but computational efficiency can be an issue when scaling to larger datasets. As its name suggests, the second method involves an examination of the log data’s statistical and numerical features and adapting a machine learning model for decision making. The use of numerical features to describe log data has significant computational efficiency improvement over working directly with sequences. The last approach adopts deep learning models that process the log data in sequential format, enabling more sophisticated feature extraction that often exceeds human capability. In this research, all three methods are implemented and evaluated in a controlled testing environment, and their strengths and weaknesses are comparatively evaluated. This study also reports on a novel finding that the time information in a log sequence plays an important role in distinguishing a faulty condition from a normal one. For most software systems, the log sequences are unevenly spaced, meaning that the timestamps associated with log data are nonuniform. Existing log analysis studies generally overlooked the time information while emphasizing log sequences. This research proposes a novel deep learning structure to unify the processing of timestamps and log sequences. The timestamps are integrated through interpolation at an intermediate layer of a neural network. Testing results demonstrate that the inclusion of timestamps makes a significant contribution to identifying a fault, and that models using time stamps can push the performance to a higher level. / Dissertation / Doctor of Engineering (DEng)

Vision-based object stabilisation is an exciting and challenging area of research, and is one that promises great technical advancements in the field of computer vision. As humans, we are capable of a tremendous array of skilful interactions, particularly when balancing unstable objects that have complex, non-linear dynamics. These complex dynamics impose a difficult control problem, since the object must be stabilised through collaboration between applied forces and vision-based feedback. To coordinate our actions and facilitate delivery of precise amounts of muscle torque, we primarily use our eyes to provide feedback in a closed-loop control scheme. This ability to control an inherently unstable object by vision-only feedback demonstrates an exceptionally high degree of voluntary motor skill. Despite the pervasiveness of vision-based stabilisation in humans and animals, relatively little is known about the neural strategies used to achieve this task. In the last few decades, with advancements in technology, we have tried to impart the skill of vision-based object stabilisation to machines, with varying degrees of success. Within the context of this research, we continue this pursuit by employing the classic Cart Inverted Pendulum; an inherently unstable, non-linear system to investigate dynamic object balancing by vision-only feedback. The Inverted Pendulum is considered to be one of the most fundamental benchmark systems in control theory; as a platform, it provides us with a strong, well established test bed for this research. We seek to discover what strategies are used to stabilise the Cart Inverted Pendulum, and to determine if these strategies can be deployed in Real-Time, using cost-effective solutions. The thesis confronts, and overcomes the problems imposed by low-bandwidth USB cameras; such as poor colour-balance, image noise and low frame rates etc., to successfully achieve vision-based stabilisation. The thesis presents a comprehensive vision-based control system that is capable of balancing an inverted pendulum with a resting oscillation of approximately ±1º. We employ a novel, segment-based location and tracking algorithm, which was found to have excellent noise immunity and enhanced robustness. We successfully demonstrate the resilience of the tracking and pose estimation algorithm against visual disturbances in Real-Time, and with minimal recovery delay. The algorithm was evaluated against peer reviewed research; in terms of processing time, amplitude of oscillation, measurement accuracy and resting oscillation. For each key performance indicator, our system was found to be superior in many cases to that found in the literature. The thesis also delivers a complete test software environment, where vision-based algorithms can be evaluated. This environment includes a flexible tracking model generator to allow customisation of visual markers used by the system. We conclude by successfully performing off-line optimization of our method by means of Artificial Neural Networks, to achieve a significant improvement in angle measurement accuracy. / Goodrich Engine Control Systems and Balfour Beatty Rail Technologies

No description available.