No description available.

This thesis addresses the crucial problem of knowledge transfer and retention in deep neural networks. The ability to transfer knowledge from previously learned tasks and retain it for future use is essential for machine learning models to continually adapt to new tasks and improve their overall performance. In principle, knowledge can be transferred between any type of task, but we believe it to be particularly challenging in the field of computer vision, where the size and diversity of visual data often result in high compute requirements and the need for large, complex models. Hence, we analyze transfer and retention learning between unsupervised and supervised visual tasks, which form the main focus of this thesis. We categorize our efforts into several knowledge transfer and retention paradigms, and we tackle them with several contributions for the scientific community. The thesis proposes settings and methods based on knowledge distillation and self-supervised learning techniques. In particular, we devise two novel continual learning settings and seven new methods for knowledge transfer and retention, setting new state-of-the-art in a wide range of tasks. In conclusion, this thesis provides a valuable contribution to the field of computer vision and machine learning and sets a foundation for future work in this area.

Despite considerable development in the therapeutic repertoire for managing cancer-related malignancies, head and neck cancer mortality has not significantly improved. The burden of HNSCC fluctuates across countries and has been associated with exposure to tobacco-derived carcinogens, excessive alcohol consumption or combinations. Due to late detection, patients often present with oral pre-malignant lesions which have progressed to an advanced stage of HNSCC. In this study, the samples were from a male cohort as generally, men are at two to four-fold higher risk than women with over 90% of HNSCCs arising in the upper aerodigestive tract. Therefore, the purpose of this thesis was to identify HNSCC biomarkers in males associated within defined anatomical region (tongue) and causative agents, specific to smoking. An iTRAQ proteomic approach was used to profile protein changes in matched normal and tumour samples from male non-smoking (n=6) and smoking patients (n=6) with tongue carcinomas revealing identification of potential targets specific to cancer. Samples were subjected to liquid nitrogen cryo-pulverisation and protein determination. Protein extracts from the same category were pooled, trypsin digested and iTRAQ 4-plex labelled. Data was generated by 2D-LC/MS on an Orbitrap Fusion and significantly changed proteins (median ± SD) were subject to bioinformatics appraisal. A total of 3426 proteins were identified and quantified by proteomic analysis. Comparison of non-smoker tumour (NS:T) with smoker tumour (S:T) distinguished 64 proteins that were upregulated and 62 downregulated, S:T vs S:N categorised 349 proteins up- and 395 down-regulated respectively and NS:T vs NS:N identified 469 proteins up- and 431 down-regulated, respectively. Arginase-1 (ARG1), Keratin Type-2 Cytoskeletal 8 (KRT8), Lipocalin-1 (LCN1) and DNA replication licensing factor MCM2 (MCM2) were identified as biologically associated with smoking compared to non-smoking, providing viable targets for verification by immunochemical methods which further supported the proteomic data. Overall, the project demonstrated the importance of using matched biopsies with good clinicopathological data for experimental design and provided a set of unique targets for a more expanded verification study.

Bistro Girls is an interconnected collection of short stories focusing on characters whose lives intertwine in the affluent Floridian town of Bellamy Park. In "Bistro Girls of Bellamy Park," a senior at a privileged college struggles to confront an old friend who has slipped into addiction. In "Bobby Blues," two women's stifling situations with live-in boyfriends give way to a small hope as Valerie casts aside her illusions and leaves to find a new apartment, temporarily freeing herself from the pattern of relying on a man. In these stories people wrestle with flawed concepts of personal identity that create outward limitations in their interactions with those they care about most. In "Disconnect," an eccentric millionaire struggles with spirituality and a romance spoiled by his inability to find satisfaction. In "The Coffee Shop," the emotionally removed Don leaves Valerie in the inevitable position to find contentment through self-reliance. Through trial and error, the obstacles of insecurity and disillusionment can at times be overcome. In "Scout's Honor," a young woman marries under the spell of fateful disillusionment, with tragic results. An annulment is the catalyst for her maturity, yet the road before her promises to be a long, painful one. As the characters come closer to acceptance of the imperfections and possibilities in themselves and the world around them, there is almost always some hope, no matter how difficult the means to get there.

A study was conducted to determine the effects of directed discovery-based teaching methods (hands-on learning) and weekly quizzes on short-term learning and long-term retention of course material in an introductory geosciences laboratory course. Assessment of learning was accomplished using percentages of correct responses to questions on two tests, using percentages from the first semester of the study as a baseline to which data from each subsequent semester were compared to determine the effects of the study variable that was introduced. Student evaluations of value, meaning, and enjoyment of the course were also investigated through the use of an essay question at the end of the second test. The study revealed that directed discovery-based methods were successful for the teaching of some subject material, but not for all, and that the method did not necessarily enhance learning of scientific vocabulary. Weekly quizzes resulted in improved learning in all subject areas. Simultaneous use of traditional and directed-discovery teaching methods as well as weekly quizzes is recommended.

Computational methods are commonly used by materials scientists to make predictions about materials. These methods can achieve in hours what would take days or weeks to accomplish in a lab. However, there are limits to what computational methods can do and how accurate the predictions are.A limiting factor for computational materials science is the size of the search space. The space of potential materials is infinite. Selecting specific systems of elements on a fixed lattice to study reduces the number of possible arrangements of atoms in the lattice to a finite number. However, this number can still be very large. Additionally this list of arrangements will contain duplicates, i.e., two different atomic arrangements could be equivalent by a rotation or translation of the lattice. Using symmetry to eliminate the duplicates saves time and resources. In order to ensure that the final list of unique structures will fit into computer memory it is also useful to know how many unique arrangements there are before actually finding them. For this reason the Pòlya enumeration algorithm was created to determine the number of unique arrangements before enumerating them. A new atomic enumeration algorithm has also been implemented in the enumlib package. This new algorithm has been optimized to find the symmetrically unique arrangements for systems with large amounts of configurational freedom, such as high-entropy alloys, which have been too computationally expensive for other algorithms.A popular computational method in materials science is Density Functional Theory (DFT). DFT codes perform first principles calculations by calculating the electron energy using numerical integrals. It is well known that the accuracy of the integrals depends heavily on the number of sample points, k-points, used. We have conducted a detailed study of how k-point sampling methods effect the accuracy of DFT calculations. This study shows that the most efficient k-point grids are those that have the fewest symmetrically distinct k-points, we call these general regular (GR) grids. GR grids are, however, difficult to generate, requiring a search across many possible grids. In order to make GR grids more accessible to the DFT community we have implemented an algorithm that can search k-point grids for the grid that has the fewest symmetry reduction in a matter of seconds.

No description available.

No description available.

No description available.

No description available.