We introduce a new, group fairness-inspired stochastic multi-armed bandit problem in the pure exploration setting. We look at the discrepancy between an arm’s mean reward from a group and the highest mean reward for any arm from that group, and call this the disappointment that group suffers from that arm. We define the optimal arm to be the one that minimizes the maximum disappointment over all groups. This optimal arm addresses one problem with maximin fairness, where the group used to choose the maximin best arm suffers little disappointment regardless of what arm is picked, but another group suffers significantly more disappointment by picking that arm as the best one. The challenge of this problem is that the highest mean reward for a group and the arm that gives that reward are unknown. This means we need to pull arms for multiple goals: to find the optimal arm, and to estimate the highest mean reward of certain groups. This leads to the new adaptive sampling algorithm for best arm identification in the fixed confidence setting called MD-LUCB, or Minimax Disappointment LUCB. We prove bounds on MD-LUCB’s sample complexity and then study its performance with empirical simulations. / Graduate

This thesis explores the complex link between Latino individuals' self-esteem and media depictions. This thesis was created with the intent to fill a current gap in communication research by providing original data and addressing the lack of quantitative research surrounding the Latino community. The Latino Perceptions of Realism in TV and Film Scale was created as a new tool to measure opinions about ethno-cultural portrayal. This thesis aims to clarify how Latino people's self-esteem is shaped by the perceived realism of media depictions, drawing on Social Identity Theory and Social Learning Theory. This thesis examines how Latinos are portrayed in movies to determine how media exposure affects how the Latino population views their self-esteem. This study uses empirical analysis to further our understanding of the relationship between media portrayal, social identity development, and self-esteem in the Latino community. The thesis offers original quantitative data using a survey-based approach. It introduces the Latino Perceptions of Realism in TV and Film Scale, a new survey tool for gauging perceptions of Latino interpretations of television and film Latino characters. Regression analysis was employed to investigate how media depictions affect the development of self-esteem in the Latino population. Key findings provide insight into the complex relationships between media portrayal and the way Latinos view their ethnic group in film and television. Demographic factors such as age and gender, the amount of media consumption, intellectual capacity, and work ethic were revealed as important determinants impacting the development of Latino self-esteem. This study contributes new data to the discipline of media effects research, extending the conversation on the stereotypical representation of Latinos in television and film. It also provides insight into how Latinos view their self-esteem based on character portrayals in film and television. / MACOM / This thesis examines the complex relationship between how Latino individuals identify and how the media portrays them. By providing new perspectives and addressing the lack of quantitative studies on the Latino population, this thesis aims to close a gap in communication research. The Latino Perceptions of Realism in TV and Film Scale is a new scale designed to gather feedback on how Latino culture is portrayed in media. Based on psychological theories, specifically Social Identity Theory and Social Learning Theory, this thesis explores how media representations affect Latino self-esteem. Through an analysis of the representations of Latinos in television and film, the study investigates how exposure to these representations affects how Latino viewers feel about themselves. This study uses empirical analysis to expand on our knowledge of the relationship between media representation, social identity, and self-esteem in the Latino community. The results provide insight into the complex connections between how Latinos view their own ethnic group in the context of film and television and how they are portrayed in the media. This study reveals a nuanced association between media consumption patterns and the self-esteem of Latinos. Latinos' identity and self-esteem are shaped by various factors, including their views of reality, the quantity of media they consume, and demographic traits. This thesis broadens the discussion on stereotyped portrayals of Latinos in the media. It offers insights into how these representations affect Latino self-esteem by bringing fresh data to the field of media effects research.

This thesis poses and solves a distribution free learning problem with consensus that arises in the study of estimation and control strategies for distributed sensor networks. Each node i for i = 1, . . . , n of the sensor network collects independent and identically distributed local measurements {z i} := {z i j}j∈N := {(x i j , yi j )}j∈N ⊆ X × Y := Z that are generated by the probability measure ρ i on Z. Each node i for i = 1, . . . , n of the network constructs a sequence of estimates {f i k }k∈N from its local measurements {z i} and from information functionals whose values are exchanged with other nodes as specified by the communication graph G for the network. The optimal estimate of the distribution free learning problem with consensus is cast as a saddle point problem which characterizes the consensus-constrained optimal estimate. This thesis introduces a two stage learning dynamic wherein local estimation is carried out via local least square approximations based on wavelet constructions and information exchange is associated with the Lagrange multipliers of the saddle point problem. Rates of convergence for the two stage learning dynamic are derived based on certain recent probabilistic bounds derived for wavelet approximation of regressor functions. / Master of Science

Programming was included in the curriculum for mathematics at highschool 2018 and to beused as a tool for problem solving in the courses mathematics 3b, mathematics in the c-track,mathematics 4 and mathematics 5. The purpose of this study is to contribute knowledge to themathematics subject in high school by investigating and describing a teaching situation whereprogramming is used as a tool for problem solving in mathematics in high school. The study'sresearch design is design research and is based on a designed activity with principles fromblock coding and Kolb's experiential learning theory that aims to answer the question of howblock coding can introduce programming as a tool for solving mathematical problems in highschool. The study was conducted in two classes with 31 and 24 students, respectively. Thestudy shows how block coding can be used to introduce students to developing the ability tosolve mathematical problems with programming. / Programmering togs med i ämnesplanen för matematik på gymnasiet 2018 och ska användassom ett verktyg för problemlösning inom kurserna matematik 3b, matematik inom c-spåret,matematik 4 och matematik 5. Syftet med studien är bidra med kunskap till matematikämnetpå gymnasiet genom att undersöka och redogöra för en undervisningssituation därprogrammering används som verktyg för problemlösning inom matematik på gymnasiet.Studiens forskningsdesign är design research och grundar sig i en designad aktivitet medprinciper från blockprogrammering och Kolbs experiential learning theory som syftar till attsvara på frågan om hur blockprogrammering kan introducera programmering som verktyg föratt lösa matematiska problem på gymnasiet. Studien genomfördes i två klasser med 31respektive 24 elever. Studien visar hur blockprogrammering kan användas för att introduceraelever till att utveckla förmågan att lösa matematiska problem med programmering.

How can one create online educational material that support and motivate students in guiding their own learning and make meaningful instructional decisions? One of the main focuses on designing e-learning is about creating an environment where learners can actively assume control and take responsibility for their own learning with little or no guidance from the tutors. This research aims to discover a new way to design learning that would cater to individual choices and preferences. The idea goes beyond learner-centred design; it is about learner control and direction. As an option, learners should be able to choose to be in the driver’s seat, to direct their own learning journey. As a starting point, this research explores the use of two educational theories - Experiential Learning Theory (ELT) and Self-Regulated Learning (SRL) theory as the underpinning instructional design for a Learner-Directed Model to support students’ online learning in both domain knowledge and meta knowledge in the subject of computer programming. One unit material from an online Introduction to Java Programming course has been redesigned based on the proposed Learner-Directed Model for the experimental design study. The study involved a total of 35 participants divided randomly into one Experimental Group and one Control Group. They were assigned to either a Learner-Directed Model (Experimental Group) or a linear model (Control Group). Pre/post tests, survey, follow-up interview as well as log file analysis were instruments used for assessing students’ domain knowledge, meta knowledge and their attitudes for their overall learning experience. Learning experience is further broken down into perceived ease of use and user satisfaction; system usability; learner experience; and perceived controllability. The results of the study have revealed that there is statistically significant difference between the survey results for the Experimental Group and the Control Group. The Experimental Group reported a higher level of overall learning experience and better attitudes in general. However, there was no statistically significant difference existing between the two groups on the domain and meta level knowledge improvement. Based on these results, I have proposed further research directions and put forward a number of recommendations and suggestions on learner-directed e-learning design.

Eftersom fertiliteten i många utvecklade länder har gått ned är det intressant att studera orsakerna bakom detta. Om inte orsakerna bara ligger i de socioekonomiska förhållandena, vilka diskuteras mest, var kan de då sökas? Denna studie använder enkätdata med avsikt att belysa om attityder till att skaffa barn påverkas av vilka familjeförhållanden man själv växt upp under samt själv upplever under vuxet liv. Synsättet bygger på idén att det finns effekter som påverkar tänkandet genom socialisering, från en generation till en annan. Genom analys av data från undersökningen YAPS studeras om det finns några empiriska samband mellan stabilitet i familjens sammansättning under uppväxten, attityder till att leva i parförhållande och attityder till att skaffa barn. Uppsatsen använder sig av regressionsanalys för att undersöka eventuella statistiskt belagda samband. Då resultaten inte visar upp något samband med uppväxtvillkoren så undersöks även ekonomiska villkor då dessa ingått som bakgrundsdata i analysen. Resultatet pekar på samband mellan synen på barn och synen på parförhållanden samt den ekonomiska situationen i familjen under uppväxten. Den attityd till barn som man uppger verkar också vara en könsfråga då kvinnor har en mer positiv syn på barn än män enligt denna studie.

Training plays a key role in bridging employer skills needs and in raising the productivity of employees and individual firms. Many large organisations have long established in-house training programmes and the means for progressive personal development for employees. However, Smaller to Medium sized Enterprises (SMEs) have not mirrored the same. SMEs are less likely to provide off the job training to employees because of operational demands. There is a lack of structured approach to training within a SME. Despite the vast array of training programmes available to SMEs, very little has been done to tailor the training to the specific learning needs of SMEs. Web Based Training Environment (WeBTiE) is a tailored online training solution specific to a SMEs’ training needs. The portal within which online training is provided is the collaborative efforts of the Employer, Employee and Training Vendor. The portal encapsulates not only the learning content, but provides the learning support necessary to guide employees through the training and reinforce work-based practises with the use of a Community of Practice. Unlike many other training models the originality of this model is that it combines Pedagogy principles in the development of the training programme, along with e-learning model, Etivities for Structured On The Job Training. The generic nature of this model allows it to be adaptable for SMEs in a variety of industries and the tailoristic feature provides the flexibility necessary to permit the adaptability. This thesis focuses on the barriers SMEs face when attempting to train employees. By determining employee learning and training needs and establishing a ‘Learning Foundation’, can we move forward with the development of a training programme commissioned for SMEs. The work in this research is underpinned by learning theories and online learning and the application of these (adapted if necessary) in the context of the day-to-day constraints typical of SMEs.

Thesis (MA)--University of Stellenbosch, 2001. / ENGLISH ABSTRACT: The construction of macro-structures. Four reading-improvement courses were offered to grade ten pupils in an attempt to improve their general reading skills and more specifically their ability to abstract the gist (macrostructure - Van Dijk and Kintsch, 1983) of texts as well as their ability to recall detail about texts. The macro-structures that learners constructed and the number of propositions recalled from a specific section of the text during a pre-test were compared to the macrostructures they formulated and the number of propositions that they recalled during a post-test. The courses were presented in Afrikaans to four of the five grade 10 classes in a Western Cape school. (The course was also presented in English to the fifth class but that course was not taken into consideration.) The courses were conducted over eight periods of approximately 40 minutes. The content of the first three periods were the same for all four courses. During the first period the learners did a reading exercise in order to determine their reading speed and reading index. During the second and third periods they were taught skimming and study skills respectively. The first course focused on questioning skills in the last five sessions. The second course was presented in order to improve learners' structuring skills. The third course was a combination of the first two courses. A conventional speed reading course was presented to learners who did the fourth course. Multiple regression analyses were conducted to determine whether (1) the construction of macro-structures and (2) the recall of propositions from the texts could be ascribed to questioning skills, structuring skills or the combination of those skills. This was done after variance in the dependant variables brought about by differences in intelligence was controlled statistically. The results of this study indicate a statistically significant improvement to construct macro-structures for the learners who completed the courses aimed at improving structuring and questioning/structuring abilities. This improvement can be ascribed to the skills that they acquired during the courses. Traditional schema theories regard schemas or schemata as pre-fabricated structures that need only to be activated by readers. In these courses, however the emphasis was placed on the formation of a structure by the reader; on the activity performed by the reader. Hence it was referred to as a structuring course. During the courses the learners were encouraged not only to "chunk" the content of the texts, but to generalize and to construct the structure of the texts. They were encouraged to use their own words or categories. Several studies indicated that an improvement in the ability to structure texts lead to an improvement in text comprehension. There are five other reasons why structuring the content of texts will facilitate the construction of macro-structures. The first is that when subjects are given the task to remember unorganized material they structure the material to be remembered spontaneously. The second is that the semantic organization of material facilitate recall and the third that normal perception is highly structured. The fourth argument is that experts make effective use of schemes. The fifth argument is that the activity of structuring forces the reader to process the text at a deeper level of processmg. Since it was assumed that top achievers are good readers who have mastered the ability to construct macro-structures efficiently, it was argued that they would benefit more from a course that aimed at improving their question-generating skills. Thus, the fact that there was not a statistically significant improvement in their ability to construct macrostructures can be attributed to the fact that they have already mastered that ability sufficiently. It is also possible to attribute the fact that there was not a statistically significant improvement to the fact that they did not master the question-generating activities. The results of this study also indicated that learners who did the structuring, questioning and questioning/structuring courses improved their ability to recall propositions from the texts in a pre-test to a post-test. Their improved ability to recall propositions, can be ascribed to the fact that detail can be inferred from a hierarchical structure since detail facts are subsumed under thematic propositions. Put differently, micro-propositions can be inferred from macro-propositions. The improvement in the ability of subjects who did the questioning course can be ascribed to the fact that questioning helps to focus attention and that the activity of questioning leads to deeper levels of processing. Three assumptions concerning macro-structures were made in this study. It was assumed that the construction of macro-structures is an automatic and integral part of the (normal) reading process. Readers do not construct macro-structures only when they are required to do so. Good readers construct better macro-structures than poor readers. / AFRIKAANSE OPSOMMING: Die konstruksie van makrostrukture. Vier leesverbeteringskursusse is vir graad 10-leerders aangebied ten einde hul leesvaardighede te probeer verbeter; meer spesifiek, hul vermoe om die kern (die makrostrukture - Van Dijk en Kintsch, 1983) van tekste te kan konstrueer en ook om proposisies van die tekste te kan herroep. Die makrostrukture wat leerders gevorrn het en die getal proposisies wat hulle kon herroep oor 'n spesifieke teksgedeelte nadat 'n leeskursus vir hulle aangebied is, is vergelyk met die makrostrukture wat hulle gevorrn het en getal proposisies wat hulle kon herroep, voordat die kursus vir hulle aangebied is. Die kursusse is in Afrikaans aangebied vir vier van die vyf graad 10-klasse van 'n skiereilandse skool. (Die kursus is ook in Engels aangebied vir die vyfde klas, maar hul resultate is nie in berekening gebring nie, aangesien die inhoud van die tekste verskil het.) Die kursusse het bestaan uit agt lesure van 40 minute. Die inhoud van die eerste drie ·lesure was dieselfde vir al vier kursusse. Tydens die eerste lesuur het die leerders 'n leesoefening gedoen en daarna het hulle hul leesspoed en leesindeks uitgewerk. Tydens die tweede en derde lesuur is vlugleesvaardighede vir die leerders aangeleer. In die eerste kursus IS die klem tydens die laaste vyf lesure geplaas op vraagstellingsvaardighede bene wens vlugleesvaardighede en studievaardighede. Leerders is geleer om kemvrae en detailvrae te forrnuleer. Tydens die tweede kursus is struktureringsvaardighede vir leerders aangeleer in die laaste vyf lesure. Die derde kursus was 'n kombinasie van die eerste twee kursusse. Tydens die vierde kursus is 'n konvensionele spoedleeskursus vir die leerders aangebied. Deur middel van meervoudige regressie-ontledings is vasgestel of vraagstelling, strukturering of die interaksie tussen vraagstelling en strukturering variansie in (1) die konstruksie van rnakrostrukture en (2) die herroep van proposisies of feite verklaar. Dit is gedoen nadat daar statisties gekontroleer is vir die variansie in die afhanklike veranderlikes wat opgewek is deur verskille in intelligensie. Uit die resultate het dit geblyk dat daar 'n statisties beduidende verbetering was by die leerders wat die strukturering-kursus gevolg het en ook by die leerders wat die vraagstelling/strukturering-kursus gevolg het om makrostrukture te konstrueer. Hierdie verbetering sou toegeskryf kon word aan die vaardighede wat hulle tydens die kursus aangeleer het. In teenstelling met skema-teoriee wat van die veronderstelling uitgaan dat skemas vaste strukture is wat slegs deur lesers geaktiveer word, is die klem in hierdie ondersoek geplaas op die aktiewe strukturering van teksinhoude. Dit is die rede waarom na die kursus verwys word as die "strukturering' -kursus. Tydens die kursus is die leerders aangemoedig om nie net die afdelings van die tekste te groepeer nie, maar om te veralgemeen; om die teksinhoude self te struktureer deur hul eie woorde te gebruik. Benewens die feit dat dit uit verskeie ander ondersoeke geblyk het dat die strukturering van tekste deur lesers hulle leesbegrip verbeter het en hulle gehelp het om meer proposisies van 'n teks te kon herroep, sou nog vyf ander argumente ook aangevoer word waarom strukturering die vorming van makrostrukture vergemaklik. Die eerste is dat toetslinge ongestruktureerde geheuemateriaal spontaan struktureer; die tweede dat semantiese organisasie die herroep daarvan vergemaklik; die derde dat ("alledaagse") persepsie hoogs gestruktureerd is en die vierde dat deskundiges skemas effektiewer benut as beginners. Die vyfde argument is dat die aktiwiteit van strukturering dieper prosessering van leesinhoude sal meebring as gewoonlees; wanneer lesers tekste struktureer, word hulle verplig om dit uitvoerig te prosesseer. Die vraagstelling-kursus is aangebied vir die leerders in die klas wat die beste presteer het, aangesien daar van die veronderstelling uitgegaan is dat hulle goeie lesers is en dat goeie lesers alreeds daartoe in staat is om goeie makrostrukture te kan konstrueer. Die feit dat diegene wat die vraagstelling-kursus gevolg het, se vermoe om rnakrostrukture te vorm, nie 'n statisties beduidende verbetering getoon het nie, sou dus moontlik daaraan toegeskryf kon word dat hulle die vaardigheid om goeie makrostrukture te vorm, alreeds in 'n groot mate bemeester het. 'n Ander moontlikheid waarom hul vermoe om makrostrukture te konstrueer, nie 'n statisties beduidende verbetering getoon het nie, is dat hulle nie die vraagstelling-aktiwiteite deeglik bemeester het rue. Uit die resultate van hierdie ondersoek het dit ook geblyk -dat daar 'n statisties beduidende verbetering was in die vermoe van leerders wat die vraagstelling-, strukturering-, asook die vraagstellinglstruktureringkursus gevolg het om na afloop van die leeskursusse meer proposisies oor die betrokke tekste te kon herroep. Die rede waarom die strukturering-kursus effektiewer herroepprestasie in die hand gewerk het, sou toegeskryf kon word aan die feit dat iemand wat 'n struktuur gekonstrueer het, detail kan herroep na aanleiding van die gekonstrueerde raarnwerk. Anders gestel, mikroproposisies sou afgelei kon word van makroproposisies. Die rede waarom die vraagstelling-kursus 'n verbetering van leerders se herroepvermoens meegebring het, sou waarskynlik daaraan toegeskryf kon word dat vraagstelling 'n leser se aandag fokus of rig. Daarby verplig die vraagstelling-aktiwiteit lesers om tekste uitvoerig te prosesseer. Hierdie vraagstelling-aktiwiteit verplig dieper prosessering van leesinhoude as die meer oppervlakkige prosessering wat tydens gewoonlees plaasvind. In hierdie ondersoek is van die verondersteIling uitgegaan dat die vorrmng van makrostrukture 'n outomatiese en integrale deel van die leesproses is. Lesers konstrueer nie net makrostrukture wanneer die leestaak dit vereis nie. Hoewel aIle lesers makrostrukture konstrueer, vorrn goeie lesers beter rnakrostrukture as swak lesers.

In an effort to address diverse needs in K-12 education, policymakers have proposed initiatives to increase the leadership opportunities for classroom teachers. Although teachers have undertaken leadership roles in schools throughout history, the notion of teacher leadership spurred by policy is in the emergent stages. As educators and leaders in various roles implement teacher leadership policy, learning must ensue at both an individual and collective level to adapt schools to a new model of leadership. This study examined how district-level teacher leader coordinators in one Iowa region came to understand their role as leaders. A case study approach illuminated the experience of these individuals as Iowa teacher leadership policy was initially implemented. Data were collected through the use of interviews, observations, and document reviews. Participants included district-level teacher leader coordinators and regional leaders supporting teacher leadership to gain insights from multiple levels of policy implementation. The findings of this study identified themes for sensemaking and leading within the teacher leader coordinator’s role. The study revealed that coordinators were guided in learning their new teacher leadership roles through both formal policy guidance and informal interactions with leaders and peers. Coordinators enacted their leadership roles in a bridge building capacity, linking other teacher leaders to resources and modeling leadership processes and actions.

A central goal of theoretical computer science is to characterize the limits of efficient computation in a variety of models. We pursue this research objective in the contexts of communication complexity and computational learning theory. In the former case, one seeks to understand which distributed computations require a significant amount of communication among the parties involved. In the latter case, one aims to rigorously explain why computers cannot master some prediction tasks or learn from past experience. While communication and learning may seem to have little in common, they turn out to be closely related, and much insight into both can be gained by studying them jointly. Such is the approach pursued in this thesis. We answer several fundamental questions in communication complexity and learning theory and in so doing discover new relations between the two topics. A consistent theme in our work is the use of analytic methods to solve the problems at hand, such as approximation theory, Fourier analysis, matrix analysis, and duality. We contribute a novel technique, the pattern matrix method, for proving lower bounds on communication. Using our method, we solve an open problem due to Krause and Pudlák (1997) on the comparative power of two well-studied circuit classes: majority circuits and constant-depth AND/OR/NOT circuits. Next, we prove that the pattern matrix method applies not only to classical communication but also to the more powerful quantum model. In particular, we contribute lower bounds for a new class of quantum communication problems, broadly subsuming the celebrated work by Razborov (2002) who used different techniques. In addition, our method has enabled considerable progress by a number of researchers in the area of multiparty communication. Second, we study unbounded-error communication, a natural model with applications to matrix analysis, circuit complexity, and learning. We obtain essentially optimal lower bounds for all symmetric functions, giving the first strong results for unbounded-error communication in years. Next, we resolve a longstanding open problem due to Babai, Frankl, and Simon (1986) on the comparative power of unbounded-error communication and alternation, showing that [mathematical equation]. The latter result also yields an unconditional, exponential lower bound for learning DNF formulas by a large class of algorithms, which explains why this central problem in computational learning theory remains open after more than 20 years of research. We establish the computational intractability of learning intersections of halfspaces, a major unresolved challenge in computational learning theory. Specifically, we obtain the first exponential, near-optimal lower bounds for the learning complexity of this problem in Kearns’ statistical query model, Valiant’s PAC model (under standard cryptographic assumptions), and various analytic models. We also prove that the intersection of even two halfspaces on {0,1}n cannot be sign-represented by a polynomial of degree less than [Theta](square root of n), which is an exponential improvement on previous lower bounds and solves an open problem due to Klivans (2002). We fully determine the relations and gaps among three key complexity measures of a communication problem: product discrepancy, sign-rank, and discrepancy. As an application, we solve an open problem due to Kushilevitz and Nisan (1997) on distributional complexity under product versus nonproduct distributions, as well as separate the communication classes PPcc and UPPcc due to Babai, Frankl, and Simon (1986). We give interpretations of our results in purely learning-theoretic terms. / text