arXiv subjects

David Joyner

Publications and source records attributed to David Joyner.

At least 19 recordsLinked to original sources

arXiv2026

When Offline Selectors Cannot Beat the Best Single Model: A Diagnostic Study on edX Dropout Prediction

Different predictors often excel on different inputs, so picking the best one per instance promises higher accuracy than committing to a single model. In practice, selectors trained from logged data routinely fail to beat the strongest single predictor. Three causes typically go unseparated before more tuning is applied: a mismatched learner, a state that does not predict which model wins, or buffer-to-deployment label shift. A three-stage diagnostic rules them out on a shared buffer. Stage~1 estimates a local ceiling on oracle recovery from $k$-NN label consistency. Stage~2 asks whether paired BC and offline-RL learners (BC, DQN, and CQL across penalty weights) reach that ceiling. Stage~3 ablates the selector state to test whether richer features would raise it. The combined verdict points to the most promising next step: tuning the learner, redesigning the state, or collecting new data. We apply it to selecting among five dropout-prediction models on edX clickstream data. Across 16 windows, the oracle beats the strongest single base model by 9.7 accuracy points on average, yet BC, DQN, and CQL land in the same test-accuracy band below it (robust to a tenfold buffer sweep and $N{=}2{,}000$ held-out examples). The bottleneck is local representational ambiguity: CQL closes the imitation gap without a deployment gain (not conservatism), regret clusters tightly across learners (not tie-breaking), and the three learners converge on test accuracy (not shift). The next iteration should change the state or collect new data, not tune the offline learner further.

cs.LG↗

arXiv2025

Using Sentiment Analysis to Investigate Peer Feedback by Native and Non-Native English Speakers

Graduate-level CS programs in the U.S. increasingly enroll international students, with 60.2 percent of master's degrees in 2023 awarded to non-U.S. students. Many of these students take online courses, where peer feedback is used to engage students and improve pedagogy in a scalable manner. Since these courses are conducted in English, many students study in a language other than their first. This paper examines how native versus non-native English speaker status affects three metrics of peer feedback experience in online U.S.-based computing courses. Using the Twitter-roBERTa-based model, we analyze the sentiment of peer reviews written by and to a random sample of 500 students. We then relate sentiment scores and peer feedback ratings to students' language background. Results show that native English speakers rate feedback less favorably, while non-native speakers write more positively but receive less positive sentiment in return. When controlling for sex and age, significant interactions emerge, suggesting that language background plays a modest but complex role in shaping peer feedback experiences.

cs.CL↗

arXiv2024

A Comparative Analysis of Student Performance Predictions in Online Courses using Heterogeneous Knowledge Graphs

As online courses become the norm in the higher-education landscape, investigations into student performance between students who take online vs on-campus versions of classes become necessary. While attention has been given to looking at differences in learning outcomes through comparisons of students' end performance, less attention has been given in comparing students' engagement patterns between different modalities. In this study, we analyze a heterogeneous knowledge graph consisting of students, course videos, formative assessments and their interactions to predict student performance via a Graph Convolutional Network (GCN). Using students' performance on the assessments, we attempt to determine a useful model for identifying at-risk students. We then compare the models generated between 5 on-campus and 2 fully-online MOOC-style instances of the same course. The model developed achieved a 70-90\% accuracy of predicting whether a student would pass a particular problem set based on content consumed, course instance, and modality.

cs.CY↗

arXiv2021

The RSNA-ASNR-MICCAI BraTS 2021 Benchmark on Brain Tumor Segmentation and Radiogenomic Classification

The BraTS 2021 challenge celebrates its 10th anniversary and is jointly organized by the Radiological Society of North America (RSNA), the American Society of Neuroradiology (ASNR), and the Medical Image Computing and Computer Assisted Interventions (MICCAI) society. Since its inception, BraTS has been focusing on being a common benchmarking venue for brain glioma segmentation algorithms, with well-curated multi-institutional multi-parametric magnetic resonance imaging (mpMRI) data. Gliomas are the most common primary malignancies of the central nervous system, with varying degrees of aggressiveness and prognosis. The RSNA-ASNR-MICCAI BraTS 2021 challenge targets the evaluation of computational algorithms assessing the same tumor compartmentalization, as well as the underlying tumor's molecular characterization, in pre-operative baseline mpMRI data from 2,040 patients. Specifically, the two tasks that BraTS 2021 focuses on are: a) the segmentation of the histologically distinct brain tumor sub-regions, and b) the classification of the tumor's O[6]-methylguanine-DNA methyltransferase (MGMT) promoter methylation status. The performance evaluation of all participating algorithms in BraTS 2021 will be conducted through the Sage Bionetworks Synapse platform (Task 1) and Kaggle (Task 2), concluding in distributing to the top ranked participants monetary awards of $60,000 collectively.

cs.CV↗

arXiv2019

On p-ary Bent Functions and Strongly Regular Graphs

Our main result is a generalized Dillon-type theorem, giving graph-theoretic conditions which guarantee that a $p$-ary function in an even number of variables is bent, for $p$ a prime number greater than 2. The key condition is that the component Cayley graphs associated to the values of the function are strongly regular, and either all of Latin square type, or all of negative Latin square type. Such a Latin or negative Latin square type bent function is regular or weakly regular, respectively. Its dual function has component Cayley graphs with the same parameters as those of the original function. We also give a criterion for bent functions involving structure constants of association schemes. We prove that if a $p$-ary function with component Cayley graphs of feasible degrees determines an amorphic association scheme, then it is bent. Since amorphic association schemes correspond to strongly regular graph decompositions of Latin or negative Latin square type, this result is equivalent to our main theorem. We show how to construct bent functions from orthogonal arrays and give some examples.

math.CO↗

arXiv2016

Self-inversive polynomials, curves, and codes

We study connections between self-inversive and self-reciprocal polynomials, reduction theory of binary forms, minimal models of curves, and formally self-dual codes. We prove that if $\mathcal X$ is a superelliptic curve defined over $\mathbb C$ and its reduced automorphism group is nontrivial or not isomorphic to a cyclic group, then we can write its equation as $y^n = f(x)$ or $y^n = x f(x)$, where $f(x)$ is a self-inversive or self-reciprocal polynomial. Moreover, we state a conjecture on the coefficients of the zeta polynomial of extremal formally self-dual codes.

math.CV↗

arXiv2014

Explorations of edge-weighted Cayley graphs and p-ary bent functions

Let f be a function mapping an n dimensional vector space over GF(p) to GF(p). When p is 2, Bernasconi et al. have shown that there is a correspondence between certain properties of f (e.g., if it is bent) and properties of its associated Cayley graph. Analogously, but much earlier, Dillon showed that f is bent if and only if the "level curves" of f had certain combinatorial properties (again, only when p is 2). The attempt is to investigate an analogous theory when p is greater than 2 using the (apparently new) combinatorial concept of a weighted partial difference set. More precisely, we try to investigate which properties of the Cayley graph of f can be characterized in terms of function-theoretic properties of f, and which function-theoretic properties of f correspond to combinatorial properties of the set of "level curves", i.e., the inverse map of f. While the natural generalizations of the Bernasconi correspondence and Dillon correspondence are not true in general, using extensive computations, we are able to determine a classification in some small cases. Our main conjecture is Conjecture 67.

math.CO↗

arXiv2010

Automorphism Groups on Tropical Curves: Some Cohomology Calculations

Let $X$ be an abstract tropical curve and let $G$ be a finite subgroup of the automorphism group of $X$. Let $D$ be a divisor on $X$ whose equivalence class is $G$-invariant. We address the following question: is there always a divisor $D'$ in the equivalence class of $D$ which is $G$-invariant? Our main result is that the answer is "yes" for all abstract tropical curves. A key step in our proof is a tropical analogue of Hilbert's Theorem 90.

math.AG↗

arXiv2009

A primer on computational group homology and cohomology

These are expanded lecture notes of a series of expository talks surveying basic aspects of group cohomology and homology. They were written for someone who has had a first course in graduate algebra but no background in cohomology. You should know the definition of a (left) module over a (non-commutative) ring, what $\zzz[G]$ is (where $G$ is a group written multiplicatively and $\zzz$ denotes the integers), and some ring theory and group theory. However, an attempt has been made to (a) keep the presentation as simple as possible, (b) either provide an explicit reference of proof of everything. Several computer algebra packages are used to illustrate the computations, though for various reasons we have focused on the free, open source packages such as GAP and SAGE.

math.GR↗

arXiv2008

On quadratic residue codes and hyperelliptic curves

A long standing problem has been to develop "good" binary linear codes to be used for error-correction. This paper investigates in some detail an attack on this problem using a connection between quadratic residue codes and hyperelliptic curves. One question which coding theory is used to attack is: Does there exist a c<2 such that, for all sufficiently large $p$ and all subsets S of GF(p), we have |X_S(GF(p))| < cp?

math.CO↗

arXiv2008

Automorphism groups of generalized Reed-Solomon codes

We look at AG codes associated to the projective line, re-examining the problem of determining their automorphism groups (originally investigated by Duer in 1987 using combinatorial techniques) using recent methods from algebraic geometry. We (re)classify those finite groups that can arise as the automorphism group of an AG code for the projective line and give an explicit description of how these groups appear. We also give examples of generalized Reed-Solomon codes with large automorphism groups G, such as G=PSL(2,q), and explicitly describe their G-module structure.

math.AG↗

arXiv2006

Group representations on Riemann-Roch spaces of some Hurwitz curves

Let q>1 denote an integer relatively prime to 2,3,7 and for which G=PSL(2,q) is a Hurwitz group for a smooth projective curve X defined over C. We compute the G-module structure of the Riemann-Roch space L(D), where D is an invariant divisor on X of positive degree. This depends on a computation of the ramification module, which we give explicitly. In particular, we obtain the decomposition of H^1(X,C) as a G-module.

math.AG↗

arXiv2005

Modular representations on some Riemann-Roch spaces of modular curves X(N)

We compute the PSL(2,N)-module structure of the Riemann-Roch space L(D), where D is an invariant non-special divisor on the modular curve X(N), with N > 5 prime. This depends on a computation of the ramification module, which we give explicitly. These results hold for characteristic p if X(N) has good reduction mod p and p does not divide the order of PSL(2,N). We give as examples the cases N=7, 11, which were also computed using GAP. Applications to AG codes associated to this curve are considered, and specific examples are computed using GAP and MAGMA.

math.AG↗

arXiv2005

Automorphism groups of some AG codes

We show that in many cases, the automorphism group of a curve and the permutation automorphism group of a corresponding AG code are the same. This generalizes a result of Wesemeyer beyond the case of planar curves.

math.AG↗

arXiv2004

A Question about Pic(X) as a G-module

Let G be a finite group acting faithfully on an irreducible non-singular projective curve defined over an algebraically closed field F. Does every G-invariant divisor class contain a G-invariant divisor? The answer depends only on G and not on the curve. We answer the same question for degree 0 divisor (classes). We investigate the question for cycles on varieties.

math.AG↗

arXiv2004

Representations of finite groups on Riemann-Roch spaces

We study the action of a finite group on the Riemann-Roch space of certain divisors on a curve. If $G$ is a finite subgroup of the automorphism group of a projective curve $X$ over an algebraically closed field and $D$ is a divisor on $X$ left stable by $G$ then we show the irreducible constituents of the natural representation of $G$ on the Riemann-Roch space $L(D)=L_X(D)$ are of dimension $\leq d$, where $d$ is the size of the smallest $G$-orbit acting on $X$. We give an example to show that this is, in general, sharp (i.e., that dimension $d$ irreducible constituents can occur). Connections with coding theory, in particular to permutation decoding of AG codes, are discussed in the last section. Many examples are included.

math.AG↗

arXiv2004

Remarks on codes from modular curves: MAGMA applications

Expository paper discussing AG or Goppa codes arising from curves, first from an abstract general perspective then turning to concrete examples associated to modular curves. We will try to explain these extremely technical ideas using a special case at a level to a typical graduate student with some background in modular forms, number theory, group theory, and algebraic geometry. Many examples using MAGMA are included.

math.NT↗

arXiv2003

Representations of finite groups on Riemann-Roch spaces, II

If G is a finite subgroup of the automorphism group of a projective curve X and D is a divisor on X stabilized by G, then under the assumption that D is nonspecial, we compute a simplified formula for the trace of the natural representation of G on Riemann-Roch space L(D).

math.AG↗