Measuring algorithmic bias in machine learning has historically focused on statistical inequalities pertaining to specific groups. However, the most common metrics (i.e., those focused on individualor group-conditione...
详细信息
ISBN:
(纸本)9798400716188
Measuring algorithmic bias in machine learning has historically focused on statistical inequalities pertaining to specific groups. However, the most common metrics (i.e., those focused on individualor group-conditioned error rates) are not currently well-suited to educational settings because they assume that each individual observation is independent from the others. this is not statistically appropriate when studying certain common educational outcomes, because such metrics cannot account for the relationship between students in classrooms or multiple observations per student across an academic year. In this paper, we present novel adaptations of algorithmic bias measurements for regression for both independent and nested data structures. Using hierarchical linear models, we rigorously measure algorithmic bias in a machine learning model of the relationship between student engagement in an intelligent tutoring system and year-end standardized test scores. We conclude that classroom-level influences had a small but significant effect on models. Examining significance with hierarchical linear models helps determine which inequalities in educational settings might be explained by small sample sizes rather than systematic differences.
Responding to recent questioning of learning Analytics (LA) as a field that is achieving its aim of understanding and optimising learning and the environments in which it occurs, this paper argues that there is a need...
详细信息
ISBN:
(纸本)9798400716188
Responding to recent questioning of learning Analytics (LA) as a field that is achieving its aim of understanding and optimising learning and the environments in which it occurs, this paper argues that there is a need to genuinely embrace the complexity of learning when considering the impact of LA. Rather than focusing upon ' optimisation ', we propose that LA should seek to understand and improve the complex socio-technical system in which it operates. We adopt a framework from systems theory to propose 12 different intervention points for learning systems, and apply it to two case studies. We conclude with an invitation to the community to critique and extend this proposed framework.
this paper explores the implementation of a wellbeing app in a Swedish Upper Secondary School. the aim is to understand how ideas of data driven school improvement underpinned by promises of artificial intelligence (A...
详细信息
ISBN:
(纸本)9783031735370;9783031735387
this paper explores the implementation of a wellbeing app in a Swedish Upper Secondary School. the aim is to understand how ideas of data driven school improvement underpinned by promises of artificial intelligence (AI) and learning analytics (LA) change the work of teachers. the study draws on video-ethnography from 17 meetings between five teachers/form tutors. the produced data is analysed using actor-network theory to focus on the various stages of the implementation process and the interactions between the learning analytics dashboard (LAD) and the teachers. To capture the complexity of the data, the empirical material is presented through cartoon-inspired illustrations grounded in a thinking through Cartoons methodology. Findings show how teachers took on new roles and responsibilities in relation to the wellbeing app, most notably the role of collecting data from students. Teachers came to act as data analysts which imposed constant negotiations and uncertainties. To address the declining engagement of students over time, a student-facing LAD was introduced. the teachers shifted their focus to motivate students to engage withtheir own data in different ways. Despite no improvements in students' response rates teachers remained committed to the app, trusting that new AI and LA functionalities would compensate unsatisfactory outcomes. In conclusion, instead of improving teachers' capacity to identify at-risk students, the wellbeing app increased teachers' workload and led to different dilemmas related to teacher-student relations and teachers' professional judgement.
this research paper seeks to present a comprehensive examination of algorithmic trading algorithms, encompassing a range of techniques that extend from supervised learning to sentiment-aware reinforcement-based tradin...
详细信息
Predicting the value of a company's stocks is the motive of stock market predictions. Today’s market use Machine learning to forecast current market values by training on their prior values. As part of the machin...
详细信息
Recent research has made strides toward fair machine learning. Relatively few datasets, however, are commonly examined to evaluate these fairness-aware algorithms, and even fewer in education domains, which can lead t...
详细信息
ISBN:
(纸本)9798400716188
Recent research has made strides toward fair machine learning. Relatively few datasets, however, are commonly examined to evaluate these fairness-aware algorithms, and even fewer in education domains, which can lead to a narrow focus on particular types of fairness issues. In this paper, we describe a novel dataset modification method that utilizes a genetic algorithm to induce many types of unfairness into datasets. Additionally, our method can generate an unfairness benchmark dataset from scratch (thus avoiding data collection in situations that might exploit marginalized populations), or modify an existing dataset used as a reference point. Our method can increase the unfairness by 156.3% on average across datasets and unfairness definitions while preserving AUC scores for models trained on the original dataset ( just 0.3% change, on average). We investigate the generalization of our method across educational datasets with different characteristics and evaluate three common unfairness mitigation algorithms. the results show that our method can generate datasets with different types of unfairness, large and small datasets, different types of features, and which affect models trained with different classifiers. Datasets generated withthis method can be used for benchmarking and testing for future research on the measurement and mitigation of algorithmic unfairness.
Additive preference representation is standard in Multiple Criteria Decision Analysis, and learning such a preference model dates back from the UTA method [11]. In this seminal work, an additive piece-wise linear mode...
详细信息
ISBN:
(纸本)9783031739026;9783031739033
Additive preference representation is standard in Multiple Criteria Decision Analysis, and learning such a preference model dates back from the UTA method [11]. In this seminal work, an additive piece-wise linear model is inferred from a learning set composed of pairwise comparisons. In this setting, the learning set is provided by a single Decision-Maker (DM), and an additive model is inferred to match the learning set. We extend this framework to the case where (i) multiple DMs with heterogeneous preferences provide part of the learning set, and (ii) the learning set is provided as a whole without knowing which DM expressed each pairwise comparison. Hence, the problem amounts to inferring a preference model for each DM and simultaneously "discovering" the segmentation of the learning set. In this paper, we show that this problem is computationally difficult. We propose a mathematical programming based resolution approach to solve this Preference learning and Segmentation problem (PLS). We also propose a heuristic to deal with large datasets. We study the performance of both algorithms through experiments using synthetic and real data.
the proceedings contain 42 papers. the topics discussed include: LLM-aided knowledge graph construction for zero-shot visual object state classification;enhancing Apple’s defect classification: insights from visible ...
ISBN:
(纸本)9798350375657
the proceedings contain 42 papers. the topics discussed include: LLM-aided knowledge graph construction for zero-shot visual object state classification;enhancing Apple’s defect classification: insights from visible spectrum and narrow spectral band imaging;analyzing emotional and topical patterns in conspiracy theory narratives: a discourse comparative study on the 2023 Hawaii wildfires;non-invasive estimation of moisture content in mushrooms using hyperspectral imaging and machine learning-based stacking regressor model;a concept drift based approach to evaluating model performance and theoretical lifespan;autism spectrum disorder prediction using machine learning classifiers;adversarial contrastive representation learning for passive Wi-Fi fingerprinting of individuals;and SAI-ChileanDiet: a multi-label food dataset with self-acquired images of the Chilean diet.
the proceedings contain 29 papers. the topics discussed include: meta GPT-based agent for enhanced phishing email detection;detecting CDN domain names based on resolution graph with PU learning;a new lattice-based str...
ISBN:
(纸本)9798400717994
the proceedings contain 29 papers. the topics discussed include: meta GPT-based agent for enhanced phishing email detection;detecting CDN domain names based on resolution graph with PU learning;a new lattice-based strong designated verifier signature scheme in the standard model;parallel compensation method for nonlinear impairments of optical fiber based on regular perturbation theory;an efficient and lightweight authenticated key agreement scheme cloud-assisted for smart agricultural monitoring systems;privacy protection in mobile crowdsensing based on local differential privacy;Argos: a detection-based method for log fusion and provenance graph compression;and a lightweight and trustworthy authentication protocol for UAV networks based on PUF.
Assessment remains a cornerstone of the educational process, with standardized testing often serving as a primary method for evaluating learning. However, as pedagogical approaches continue to evolve, Outcome-Based As...
详细信息
暂无评论