京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第21390号 / 情博第676号 / 新制||情||117(附属図書館) / 京都大学大学院情報学研究科知能情報学専攻 / (主査)教授 鹿島 久嗣, 教授 山本 章博, 教授 下平 英寿 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

There exist many real-world applications such as recommendation systems, document retrieval, and computational biology where the correct ordering of instances is of equal or greater importance than predicting the exact value of some discrete or continuous outcome. Learning-to-Rank (LTR) refers to a group of algorithms that apply machine learning techniques to tackle these ranking problems. Despite their empirical success, most existing LTR models are not built to be robust to errors in labeling or annotation, distributional data shift, or adversarial data perturbations. To fill this gap, we develop four LTR frameworks that are robust to various types of perturbations. First, Pairwise Elastic Net Regression Ranking (PENRR) is an elastic-net-based regression method for drug sensitivity prediction. PENRR infers robust predictors of drug responses from patient genomic information. The special design of this model (comparing each drug with other drugs in the same cell line and comparing that drug with itself in other cell lines) significantly enhances the accuracy of the drug prediction model under limited data. This approach is also able to solve the problem of fitting on the insensitive drugs that is commonly encountered in regression-based models. Second, Regression-based Ranking by Pairwise Cluster Comparisons (RRPCC) is a ridge-regression-based method for ranking clusters of similar protein complex conformations generated by an underlying docking program (i.e., ClusPro). Rather than using regression to predict scores, which would equally penalize deviations for either low-quality and high-quality clusters, we seek to predict the difference of scores for any pair of clusters corresponding to the same complex. RRPCC combines these pairwise assessments to form a ranked list of clusters, from higher to lower quality. We apply RRPCC to clusters produced by the automated docking server ClusPro and, depending on the training/validation strategy, we show. improvement by 24%–100% in ranking acceptable or better quality clusters first, and by 15%–100% in ranking medium or better quality clusters first. Third, Distributionally Robust Multi-Output Regression Ranking (DRMRR) is a listwise LTR model that induces robustness into LTR problems using the Distributionally Robust Optimization framework. Contrasting to existing methods, the scoring function of DRMRR was designed as a multivariate mapping from a feature vector to a vector of deviation scores, which captures local context information and cross-document interactions. DRMRR employs ranking metrics (i.e., NDCG) in its output. Particularly, we used the notion of position deviation to define a vector of relevance score instead of a scalar one. We then adopted the DRO framework to minimize a worst-case expected multi-output loss function over a probabilistic ambiguity set that is defined by the Wasserstein metric. We also presented an equivalent convex reformulation of the DRO problem, which is shown to be tighter than the ones proposed by the previous studies. Fourth, Inversion Transformer-based Neural Ranking (ITNR) is a Transformer-based model to predict drug responses using RNAseq gene expression profiles, drug descriptors, and drug fingerprints. It utilizes a Context-Aware-Transformer architecture as its scoring function that ensures the modeling of inter-item dependencies. We also introduced a new loss function using the concept of Inversion and approximate permutation matrices. The accuracy and robustness of these LTR models are verified through three medical applications, namely cluster ranking in protein-protein docking, medical document retrieval, and drug response prediction.

This thesis develops methods of task localization, task similarity discovery, and task transfer for eventual use in a reinforcement learning task library system, which can effectively “learn to learn,” improving its performance as it encounters various tasks over the lifetime of the learning system.

No description available.

Att skriva för hand är en viktig förmåga som används frekvent, både i skolan och i vardagen. Denna kunskapsöversikt syftar till att undersöka aktuell forskning kring skrivundervisning med särskilt intresse för handskrift. Kunskapsöversikten utgår ifrån forskningsfrågan: Vad säger forskningen om hur lärare bör undervisa för att optimera handskriftsutveckling? Genom en tematisk analys har resultatet från tio olika forskningsartiklar sammanställts och analyserats. Artiklarna har valts ut genom en litteratursökning. Resultatet visar att elever behöver få explicit undervisning och träning i skolan för att utveckla och bemästra handskrift. Vidare har handskrift visat sig ha positiva kopplingar till andra literacyrelaterade förmågor, såsom stavning och läsning. Att skriva för hand kräver både motorisk och kognitiv ansträngning och forskningen visar på vikten av att tidigt automatisera handskrift, för att kunna frigöra kognitiva resurser och underlätta för elever att fokusera på textinnehåll. Forskningen pekar även på fördelarna med att använda olika typer av verktyg i skrivundervisning, såsom olika pennor och papper samt digitala verktyg. Kunskapsöversikten har synliggjort vikten av tydlig undervisning i handskrift i lågstadiet, då en automatiserad handskriftsförmåga underlättar literacyutveckling. Därtill ges konkreta exempel på undervisningsmetoder samt både analoga och digitala verktyg som kan stödja lärare i tidig skrivundervisning.

California needs high quality teachers, particularly in schools that are located where well- prepared teachers who are committed to teaching urban youth are in short supply. Only 15–18% of traditional teacher candidates state a preference for urban settings. In contrast, the percentage of interns who state that they would prefer to teach in an urban school is 70%. Because of its ability to produce teachers willing to teach in urban schools, the California District Intern Program has been able to help alleviate the shortage of teachers willing to teach in urban settings. A key feature of district intern programs is the requirement to establish cohort structures within each program. The purpose of this study was to describe cohort support as it exists in district intern programs currently in operation in California. This included understanding what effect, if any, cohort participation has on interns' sense of personal teaching efficacy and determining to what extent the relationships formed within the cohorts provide support in both teaching and non-teaching contexts. Additionally, this study provides insight into practice and offers recommendations for improving the cohort system in district intern programs. California district interns affirm the need for cohort groups in learning to teach. There is strong agreement that participation in a cohort is a positive experience and seen by interns as being essential to their success within district intern programs. Additional analysis provided evidence that interns participation in cohort activities specifically tied to reflection is linked to a higher sense of personal teaching efficacy. This is crucial information as there is a direct relationship between teaching efficacy and higher student achievement. Regardless of whether internships exist as a result of a teacher shortage in California or because intern programs are seen as a high quality program for preparing teachers, these novices are expected to learn to teach on the job. There is clear evidence that participation in cohort groups provide interns with the support they feel is necessary for them to be successful in this endeavor. Current programs provide ample opportunities for this to occur, new programs are encouraged to provide the same variety.

No description available.

No description available.

No description available.

No description available.