<p>Computer Vision is a mathematical tool formulated to extend human vision to machines. This tool can perform various tasks such as object classification, object tracking, motion estimation, and image segmentation. These tasks find their use in many applications, namely robotics, self-driving cars, augmented reality, and mobile applications. However, opposed to the traditional technique of incorporating handcrafted features to understand images, convolution neural networks are being used to perform the same function. Computer vision applications widely use CNNs due to their stellar performance in interpreting images. Over the years, there have been numerous advancements in machine learning, particularly to CNNs. However, the need to improve their accuracy, model size and complexity increased, making their deployment in restricted environments a challenge. Many researchers proposed techniques to reduce the size of CNN while still retaining its accuracy. Few of these include network quantization, pruning, low rank, and sparse decomposition and knowledge distillation. Some methods developed efficient models from scratch. This thesis achieves a similar goal using design space exploration techniques on the latest variant of MobileNets, MobileNet V3. Using Depthwise Pointwise Depthwise (DPD) blocks, escalation in the number of expansion filters in some layers and mish activation function MobileNet V3 is reduced to 84.96% in size and made 0.2% more accurate. Furthermore, it is deployed in NXP i.MX RT1060 for image classification on CIFAR-10 dataset.</p>

This mini-dissertation aims to explore the daily realities faced by caregivers of children with autism spectrum disorder and the implications thereof on the use of a cell phone application that can assist in easing the tension between the need for support and the lack of resources to secure that support whenever necessary. Bronfenbrenner’s bio-ecological systems theory, in addition to the assets-based approach, forms the basis on which the study is conceptualised. Using a qualitative approach, a case study research design was used to select the participant by means of purposive sampling. The research participant used the application that was selected for a period of ten days, after which a semi-structured interview was conducted as the primary mode of data collection. The data were then analysed, using inductive thematic analysis, after which themes and subthemes were derived. The findings of the study indicated that the research participant faces a meaningful lack of financial support and social support, and that the cell phone application was seen as a useful tool for supporting her child in spite of this. The study contributed to expanding the current research body on this topic. A major limitation was the sample size of the study being too small for the results to be generalised. Recommendations include that cell phone applications be designed with users’ cultural and language preferences in mind and that more studies of this nature be done. / Dissertation (MEd)--University of Pretoria, 2019. / Educational Psychology / MEd / Unrestricted

Indiana University-Purdue University Indianapolis (IUPUI) / Computer Vision is a mathematical tool formulated to extend human vision to machines. This tool can perform various tasks such as object classification, object tracking, motion estimation, and image segmentation. These tasks find their use in many applications, namely robotics, self-driving cars, augmented reality, and mobile applications. However, opposed to the traditional technique of incorporating handcrafted features to understand images, convolution neural networks are being used to perform the same function. Computer vision applications widely use CNNs due to their stellar performance in interpreting images. Over the years, there have been numerous advancements in machine learning, particularly to CNNs.However, the need to improve their accuracy, model size and complexity increased, making their deployment in restricted environments a challenge. Many researchers proposed techniques to reduce the size of CNN while still retaining its accuracy. Few of these include network quantization, pruning, low rank, and sparse decomposition and knowledge distillation. Some methods developed efficient models from scratch. This thesis achieves a similar goal using design space exploration techniques on the latest variant of MobileNets, MobileNet V3. Using DPD blocks, escalation in the number of expansion filters in some layers and mish activation function MobileNet V3 is reduced to 84.96% in size and made 0.2% more accurate. Furthermore, it is deployed in NXP i.MX RT1060 for image classification on CIFAR-10 dataset.

Indiana University-Purdue University Indianapolis (IUPUI) / Component-Based Software Development (CSBD) has gained widespread importance in recent times, due to its wide-scale applicability in software development. System developers can now pick and choose from the pre-existing components to suit their requirements in order to build their system. For the purpose of developing a quality-aware system, finding the suitable components offering services is an essential and critical step. Hence, Service Discovery is an important step in the development of systems composed from already existing quality-aware software services. Currently, there is a plethora of new-age devices, such as PDAs, and cell phones that automate daily activities and provide a pervasive connectivity to users. The special characteristics of these devices (e.g., mobility, heterogeneity) make them as attractive choices to host services. Hence, they need to be considered and integrated in the service discovery process. However, due to their limitations of battery life, intermittent connectivity and processing capabilities this task is not a simple one. This research addresses this challenge of including resource constrained devices by enhancing the UniFrame Resource Discovery System (URDS) architecture. This enhanced architecture is called Mobile Device Enabled Service Discovery System (MDE-URDS). The experimental validation of the MDE-URDS suggests that it is a scalable and quality-aware system, handling the limitations of mobile devices using existing and well established algorithms and protocols such as Mobile IP.

The main goal of a scheduling process is to decide when and how to execute each of the project’s activities. Despite large variety of researched scheduling problems, the majority of them can be described as generalisations of the resource-constrained project scheduling problem (RCPSP). Because of wide applicability and challenging difficulty, RCPSP has attracted vast amount of attention in the research community and great variety of heuristics have been adapted for solving it. Even though these heuristics are structurally different and operate according to diverse principles, they are designed to obtain only one solution at a time. In the recent researches on RCPSPs, it was proven that these kind of problems have complex multimodal fitness landscapes, which are characterised by a wide solution search spaces and presence of multiple local and global optima. The main goal of this thesis is twofold. Firstly, it presents a variation of the RCPSP that considers optimisation of projects in an uncertain environment where resources are modelled to adapt to their environment and, as the result of this, improve their efficiency. Secondly, modification of a novel evolutionary computation method Cuckoo Search (CS) is proposed, which has been adapted for solving combinatorial optimisation problems and modified to obtain multiple solutions. To test the proposed methodology, two sets of experiments are carried out. Firstly, the developed algorithm is applied to a real-life software development project. Secondly, the performance of the algorithm is tested on universal benchmark instances for scheduling problems which were modified to take into account specifics of the proposed optimisation model. The results of both experiments demonstrate that the proposed methodology achieves competitive level of performance and is capable of finding multiple global solutions, as well as prove its applicability in real-life projects.

Project managers are faced with the challenging task of managing an environment filled with uncertainties that may lead to multiple disruptions during project execution. In particular, they are frequently confronted with planning for routine and non-routine unplanned work: known, identified, tasks that may or may not occur depending upon various, often unpredictable, factors. This problem is known as the stochastic task insertion problem, where tasks of deterministic duration occur stochastically. Traditionally, project managers may include an extra margin within deterministic task times or an extra time buffer may be allotted at the end of the project schedule to protect the final project completion milestone. Little scientific guidance is available to better integrate buffers strategically into the project schedule. Motivated by the Critical Chain and Buffer Management approach of Goldratt, this research identifies, defines, and demonstrates new buffer sizing techniques to improve project duration and stability metrics associated with the stochastic resource constrained project scheduling problem with stochastic task insertions. Specifically, this research defines and compares partial buffer sizing strategies for projects with varying levels of resource and network complexity factors as well as the level and location of the stochastically occurring tasks. Several project metrics may be impacted by the stochastic occurrence or non-occurrence of a task such as the project makespan and the project stability. New duration and stability metrics are developed in this research and are used to evaluate the effectiveness of the proposed buffer sizing techniques. These "robustness measures" are computed through the comparison of the characteristics of the initial schedule (termed the infeasible base schedule), a modified base schedule (or as-run schedule) and an optimized version of the base schedule (or perfect knowledge schedule). Seven new buffer sizing techniques are introduced in this research. Three are based on a fixed percentage of task duration and the remaining four provide variable buffer sizes based upon the location of the stochastic task in the schedule and knowledge of the task stochasticity characteristic. Experimental analysis shows that partial buffering produces improvements in the project stability and duration metrics when compared to other baseline scheduling approaches. Three of the new partial buffering techniques produced improvements in project metrics. One of these partial buffers was based on a fixed percentage of task duration and the other two used a variable buffer size based on knowledge of the location of the task in the project network. This research provides project schedulers with new partial buffering techniques and recommendations for the type of partial buffering technique that should be utilized when project duration and stability performance improvements are desired. When a project scheduler can identify potential unplanned work and where it might occur, the use of these partial buffer techniques will yield a better estimated makespan. Furthermore, it will result in less disruption to the planned schedule and minimize the amount of time that specific tasks will have to move to accommodate the unplanned tasks.

The area of focus for this research is the Stochastic Resource Constrained Project Scheduling Problem (SRCPSP) with Stochastic Task Insertion (STI). The STI problem is a specific form of the SRCPSP, which may be considered to be a cross between two types of problems in the general form: the Stochastic Project Scheduling Problem, and the Resource Constrained Project Scheduling Problem. The stochastic nature of this problem is in the occurrence/non-occurrence of tasks with deterministic duration. Researchers Selim (2002) and Grey (2007) laid the groundwork for the research on this problem. Selim (2002) developed a set of robustness metrics and used these to evaluate two initial baseline (predictive) scheduling techniques, optimistic (0% buffer) and pessimistic (100% buffer), where none or all of the stochastic tasks were scheduled, respectively. Grey (2007) expanded the research by developing a new partial buffering strategy for the initial baseline predictive schedule for this problem and found the partial buffering strategy to be superior to Selim s extreme buffering approach. The current research continues this work by focusing on resource aspects of the problem, new buffering approaches, and a new rescheduling method. If resource usage is important to project managers, then a set of metrics that describes changes to the resource flow would be important to measure between the initial baseline predictive schedule and the final as-run schedule. Two new sets of resource metrics were constructed regarding resource utilization and resource flow. Using these new metrics, as well as the Selim/Grey metrics, a new buffering approach was developed that used resource information to size the buffers. The resource-sized buffers did not show to have significant improvement over Grey s 50% buffer used as a benchmark. The new resource metrics were used to validate that the 50% buffering strategy is superior to the 0% or 100% buffering by Selim. Recognizing that partial buffers appear to be the most promising initial baseline development approach for STI problems, and understanding that experienced project managers may be able to predict stochastic probabilities based on prior projects, the next phase of the research developed a new set of buffering strategies where buffers are inserted that are proportional to the probability of occurrence. The results of this proportional buffering strategy were very positive, with the majority of the metrics (both robustness and resource), except for stability metrics, improved by using the proportional buffer. Finally, it was recognized that all research thus far for the SRCPSP with STI focused solely on the development of predictive schedules. Therefore, the final phase of this research developed a new reactive strategy that tested three different rescheduling points during schedule eventuation when a complete rescheduling of the latter portion of the schedule would occur. The results of this new reactive technique indicate that rescheduling improves the schedule performance in only a few metrics under very specific network characteristics (those networks with the least restrictive parameters). This research was conducted with extensive use of Base SAS v9.2 combined with SAS/OR procedures to solve project networks, solve resource flow problems, and implement reactive scheduling heuristics. Additionally, Base SAS code was paired with Visual Basic for Applications in Excel 2003 to implement an automated Gantt chart generator that provided visual inspection for validation of the repair heuristics. The results of this research when combined with the results of Selim and Grey provide strong guidance for project managers regarding how to develop baseline predictive schedules and how to reschedule the project as stochastic tasks (e.g. unplanned work) do or do not occur. Specifically, the results and recommendations are provided in a summary tabular format that describes the recommended initial baseline development approach if a project manager has a good idea of the level and location of the stochasticity for the network, highlights two cases where rescheduling during schedule eventuation may be beneficial, and shows when buffering proportional to the probability of occurrence is recommended, or not recommended, or the cases where the evidence is inconclusive.

Yes / An Improved Discrete Cuckoo Search (IDCS) is proposed in this paper to solve resource-constrained project scheduling problems (RCPSPs). The original Cuckoo Search (CS) was inspired by the breeding behaviour of some cuckoo species and was designed specifically for application in continuous optimisation problems, in which the algorithm had been demonstrated to be effective. The proposed IDCS aims to improve the original CS for solving discrete scheduling problems by reinterpreting its key elements: solution representation scheme, Lévy flight and solution improvement operators. An event list solution representation scheme has been used to present projects and a novel event movement and an event recombination operator has been developed to ensure better quality of received results and improve the efficiency of the algorithm. Numerical results have demonstrated that the proposed IDCS can achieve a competitive level of performance compared to other state-of-the-art metaheuristics in solving a set of benchmark instances from a well-known PSPLIB library, especially in solving complex benchmark instances. / Partially funded by the Innovate UK project HARNET – Harmonised Antennas, Radios and Networks under contract no. 100004607.

This research presents a Resource-constrained Critical Path Method (RCPM) technique that capitalizes on and improves the existing Critical Path Method (CPM) and Resource-Constrained Scheduling (RCS) techniques. A traditional CPM schedule is not realistic since it assumes unlimited resources, some of which are highly limited in practice. Although traditional RCS techniques can consider resource limitations, they do not provide correct floats and the critical path as the CPM does. The difference between the theoretical remaining total float and the real remaining total float is referred to as "Phantom Float" in this study. Another disadvantage of the traditional RCS techniques is that work sequence in the schedule could be considerably changed with a schedule update resulting in high costs to reorganize it. These problems are caused by the fact that, in addition to technological relationships, a resource-constrained schedule contains resource dependencies between activities that are neglected in traditional RCS techniques. This study proposes a step-by-step RCPM algorithm to consider those resource-constrained relationships. Hence, the method can identify real floats and correct critical paths, considering both technological and resource-dependent relationships. RCPM also provides a certain level of stability with a schedule update due to the newly identified resource relationships. Based on the RCPM algorithm, a prototype RCPM system has been developed using Visual C++, Visual Basic, and Ra (Primavera Project Planner API). The system is integrated with P3, so that it reads project information directly from a P3 project, performs necessary RCPM procedures, and updates the P3 project to contain identified resource relationships. To make the system more practical, functions to handle multiple project calendars and progressed schedules have been included as well. / Ph. D.

Yes / In this paper, a new population-based and nature-inspired metaheuristic algorithm, Discrete Flower Pollination Algorithm (DFPA), is presented to solve the Resource Constrained Project Scheduling Problem (RCPSP). The DFPA is a modification of existing Flower Pollination Algorithm adapted for solving combinatorial optimization problems by changing some of the algorithm's core concepts, such as flower, global pollination, Lévy flight, local pollination. The proposed DFPA is then tested on sets of benchmark instances and its performance is compared against other existing metaheuristic algorithms. The numerical results have shown that the proposed algorithm is efficient and outperforms several other popular metaheuristic algorithms, both in terms of quality of the results and execution time. Being discrete, the proposed algorithm can be used to solve any other combinatorial optimization problems. / Innovate UK / Awarded 'Best paper of the Month'