<p>Though modern deep learning based approaches have achieved remarkable progress in computer vision community such as image classification using a static image dataset, it suf- fers from catastrophic forgetting when learning new classes incrementally in a phase-by-phase fashion, in which only data for new classes are provided at each learning phase. In this work we focus on continual learning with the objective of learning new tasks from sequentially available data without forgetting the learned knowledge. We study this problem from three perspectives including (1) continual learning in online scenario where each data is used only once for training (2) continual learning in unsupervised scenario where no class label is pro- vided and (3) continual learning in real world applications. Specifically, for problem (1), we proposed a variant of knowledge distillation loss together with a two-step learning technique to efficiently maintain the learned knowledge and a novel candidates selection algorithm to reduce the prediction bias towards new classes. For problem (2), we introduced a new framework for unsupervised continual learning by using pseudo labels obtained from cluster assignments and an efficient out-of-distribution detector is designed to identify whether each new data belongs to new or learned classes in unsupervised scenario. For problem (3), we proposed a novel training regime targeted on food images using balanced training batch and a more efficient exemplar selection algorithm. Besides, we further proposed an exemplar-free continual learning approach to address the memory issue and privacy concerns caused by storing part of old data as exemplars.</p> <p>In addition to the work related to continual learning, we study the image-based dietary assessment with the objective of determining what someone eats and how much energy is consumed during the course of a day by using food or eating scene images. Specifically, we proposed a multi-task framework for simultaneously classification and portion size estima- tion by future fusion and soft-parameter sharing between backbone networks. Besides, we introduce RGB-Distribution image by concatenating the RGB image with the energy distri- bution map as the fourth channel, which is then used for end-to-end multi-food recognition and portion size estimation.</p>

<p>Previous studies have reported inconsistent results for people with Autism Spectrum Disorders (ASD) in both biological motion and Navon tasks. Each of these tasks require both local and global processing and were used in this study to compare processing styles of both ASD and typical groups. In the biological motion study, the ASD and the typical groups completed an emotion and a direction-discrimination experiment with happy and angry point-light walkers, which were presented in four different stimulus conditions: upright, inverted, scrambled and random. Overall, the ASD group had higher reaction times and lower accuracy, but the effect of condition did not differ between groups. Both groups performed worse in terms of accuracy and reaction times in the scrambled (i.e., local information only) conditions, therefore revealing a global bias in the processing of biological motion information. In the Navon task study with the same participants, typical individuals exhibited a global precedence effect, manifested as lower reaction times for global stimuli as well as global interference in “look for only local digits” task. However, individuals in the ASD group did not, on average, show a local or a global bias. In a subsequent analysis, the ASD group was divided into locally-biased and globally-biased sub-groups. Now, when a three way analysis between typical and the two ASD groups was performed, the globally-biased group’s performance was not distinguishable from that of the typical group, while no global bias was observed for the locally-biased group. When these two groups were compared on the Biological Motion study, the locally-biased group had no reaction time difference across conditions including both biological motion and Navon tasks, unlike the globally-biased group, who displayed higher reaction times for the scrambled condition, just like the typical group did. Therefore, it is possible that the inconsistencies in the local-global processing literature of individuals with ASD may have resulted because the studies did not account for individual differences in processing styles within the ASD groups that may be variable, unlike typical individuals who have a global bias for most tasks.</p> / Master of Science (MSc)

<p> Increasing evidence indicates that giving individuals control over their practice schedule facilitates motor learning (Titzer, Shea, & Romack, 1993; Wu & Magill, 2004, 2005). A recent study within cognitive psychology (Son, 2004) found that individuals massed practice on tasks they judged as hard but spaced practice on tasks they judged as easy. The purpose of this experiment was to examine how self-regulated practice impacts motor learning and the strategies used by individuals as a function of task complexity. Participants were required to move a mouse-driven cursor through a pattern of coloured squares, pausing only long enough in each square to make an appropriate button press (white square=left button, black square=right button). Task complexity (4 easy and 4 hard patterns) was determined by the combined effects of the arrangement of the grid of squares and the hand used to manipulate the mouse (easy =dominant hand, hard=non-dominant hand). Participants were randomly assigned to one of eight groups: blocked, random, self-regulated, and yoked to self-regulated, performing either the easy or hard tasks. The number of switches between patterns were recorded as well as temporal and accuracy measures. The self-regulated groups were ineffective in acquisition, but showed the most stable and improved performance in retention, irrespective of task difficulty. Moreover, although switch strategies of the self-regulated groups differed between and within task complexity, the motor learning advantage was generalized. Taken together, these results reveal that an individual's strategic approach to practice may change as a function of task complexity, with no detriment to motor learning and adds to the growing body of literature that suggests self-regulated practice is an important variable for effective motor learning.</p> / Thesis / Master of Science (MSc)

The daily, weekly, and unscheduled tasks for school administrators have increased in number and scope over the years, however surprisingly little is known about what principals do on a day-to-day basis and how this varies across schools. Since the effect of principal leadership behaviors, specifically how principals manage their time to accomplish important tasks, is one key to the success of schools, it is important to understand what effective principals do to accomplish this. The purpose of this study was to find out what the differences are in how principals in high and low-performing middle schools spend their time and to determine what relationships exist between the principal's time-on-tasks and school performance. In the literature review, the researcher identified seven categories of time use to collect and classify time-on-tasks data. The categories include: (a) administration/operations, (b) organization management, (c) day-to-day instruction, (d) instructional program, (e) internal relations, (f) external relations and (g) other (Horng, Klasik, & Loeb, 2010). The researcher collected time-on-tasks data from principals of high and low-performing middle schools in Virginia and analyzed the data to determine what relationships exist between the principal's time-on-tasks and school performance. Data analyses revealed that there are significant differences in the amount of time principals at high-performing schools devote to each of the time-on-tasks categories, as compared to the amount of time allocated by their counterparts at low-performing schools. In this study, principals as a whole and principals in the high-performing subgroup spend the largest percentage of time on tasks related to administration and operations, while principals in the low-performing subgroup spend the largest percentage of time on day-to-day instruction. Data also suggest that time spent on tasks related to internal relations is positively correlated with student performance on mathematics and reading tests. When demographic factors are combined with the time-on-tasks categories, a regression analysis suggests that the strongest contributing factor to mathematics and reading test scores is the socioeconomic status of the school with a strong negative correlation between the percentage of students on free/reduced lunch and test scores for mathematics and reading. / Ed. D.

This study explores self-efficacy as an explanation for individual differences in participation in a manufacturing organization with a structured participation program. Participation covers three distinct dimensions of behavior: (1) decision-making pertaining to tasks, (2) good citizenship in the form of extra effort and helping others, and (3) contributing to improvement in work processes. Self-efficacy refers to an individual's belief that he/she can successfully perform an activity in a specific situation. The project was based on an action research design in which the first phase examined the dimensionality of participation self-efficacy, the relationship between participation self-efficacy and actual ratings of participation, and the relationship between perceptions of situational factors and self-efficacy. Exploratory factor analysis found preliminary support for the three proposed dimensions of participation self-efficacy and also for a fourth communication dimension. Weak (e.g., r=.27) and non-significant correlations were found between self-efficacy and actual participation ratings. Situational factors were examined as perceptions of barriers which were proposed to be inversely related to self-efficacy. As expected, negative correlations (ranging from -.28 to -.45) were found between perceptions of situational factors and participation self-efficacy. Phase two of the project evaluated a critical thinking training program designed to increase employee participation. A Pre-test Post-test! Nonequivalent control group design was used to study the influence of training on learning self-efficacy and participation self-efficacy. ANOYA and ANCOVA found no significant differences in post-training self-efficacy between the trained group and the control group. / Ph. D.

There have been thousands of studies on teams and their performance, but there are still many unanswered questions. An important one is how an organization's reward structure supports the growing trend of using teams. Many organizations implement teams without changing the organizational systems to align with and support the use of teams, i.e., training, feedback, information and reward systems. As predicted by many authorities in the field of team effectiveness research, these teams often fail. One organizational subsystem that has been determined to be important is the reward structure. If the reward structure is not changed to support a team-based structure, the misalignment could negatively impact team effectiveness. This research investigated the relationship between reward structure and team effectiveness using a laboratory experiment. This experiment involved groups of students working as a team on a design problem. The independent variable is the type of reward structure, manipulated over three levels: interdependent (group), independent (individual) and mixed rewards (both group and individual). The experiment used a design task, intended to be more representative of project team work where team members were assigned a functional discipline and worked together to solve a design problem. The primary dependent variable in this study was team effectiveness: team performance as measured by the quality of the team's design, satisfaction of team members, and the ability and desire of team members to work together in the future. Other control variables investigated for their effect on these dependent variables included: cooperative behaviors, reward valence, effort, and autonomy preferences. Few significant effects of reward structure were found. The reward treatment had a significant main effect on both cooperation and effort, but little difference existed between reward treatments. Some unusual results were found in the relationship between effort and cooperation with performance. Both effort and cooperation were negatively related to team performance. Cooperation, satisfaction and ability to exist were all found to be correlated. No one reward structure was found to be significantly better than any of the others in terms of team effectiveness or team process. / Master of Science

English classrooms in Sweden host students of mixed abilities, all entitled to adapted education. Despite this, English language teaching, unlike other subjects, primarily relies on course books, serving as both learning tools and syllabus references. However, previous research has highlighted shortcomings in these books, such as insufficient vocabulary exercises and gaps in alignment with the syllabus. Although research and the Swedish National Agency of Education advocate for less reliance on course books and more communicative teaching, many teachers still find them useful. This study aims to explore the extent to which these course books allow for differentiation, to meet the needs of all students. The study employed triangulation method, by mixing a content analysis of 15 course books and interviews with six English teachers. The investigation's primary conclusions are that there are not many opportunities for differentiation within the books, but teachers seem to make do with alternative resources, adapting the books tasks and giving students varying degrees of support. While some differentiation is possible within course books, it may not fully address the needs of students of mixed-ability classrooms. Therefore, it could be concluded that the current approach may be insufficient to ensure that every student's needs are met, potentially limiting their language acquisition opportunities.

Previous research examining the impact of training and practice effects on modifying performance of fluid intelligence tasks (Gf) and crystallized intelligence tasks (Gc) were extended to include self-rated performance attributions among non-depressed and depressed older persons. The following general questions were addressed. How does level of depression affect performance on Gf and Gc measures and performance attributions? How does level of depression and degree of benefit from either training or practice relate to changes in attributional styles? The framework used for predicting shifts in attributional styles was the reformulated learned helplessness model. Three hundred twenty-five community-dwelling older persons completed the Gf/Gc Sampler, Beck Depression Inventory, and Attributions for Success/Failure Questionnaire at pretest, posttest (one week), and follow-up (one month). Between the pretest and posttest sessions, subjects participated in one of three experimental conditions; (a) cognitive (induction) training, (b) stress inoculation training, and (c) no-contact control groups. The results from univariate and multivariate analysis of covariance procedures provided partial support for the hypotheses. At pretest, both non-depressed and depressed older persons had internal attributional styles, although based on differential performance outcomes. The depressed persons were found to have more failure experiences as a result of their significantly poorer performance on Gf tasks, versus the non-depressed. Specific Gf training effects were documented regarding attributional shifts for the non-depressed, while there were no changes on their attributional style due to practice on either Gf or Gc tasks. In contrast, only differential practice effects were documented for depressed subjects across Gf and Gc tasks. The importance of assessing personality dimensions in older persons and their xelationship to training and practice effects were discussed, in addition to limitations of the study and suggestions for future research.

No description available.

The aim of this thesis is to create a collection of tasks that would address the students of secondary schools, secondary vocational schools and grammar schools by using a spreadsheet. Collection of tasks will respond to the theoretical background, which is mainly based on the requirements for the knowledge and skills to work with spreadsheets in the curricula of the Czech and Slovak secondary schools and ECDL standards. Other theoretical bases are the analysis of learning tasks in the books, a survey among secondary school teachers, through which we discover how the spreadsheet is used and what tasks the students dealt with this application. Collection of tasks will offer a wide range of tasks of varying difficulty allowing cross-curricular use. Powered by TCPDF (www.tcpdf.org)