arXiv subjects

Emily Dodwell

Publications and source records attributed to Emily Dodwell.

6 recordsLinked to original sources

arXiv2026

Reforming the Mechanism: Editing Reasoning Patterns in LLMs with Circuit Reshaping

Large language models (LLMs) often exhibit flawed reasoning ability that undermines reliability. Existing approaches to improving reasoning typically treat it as a general and monolithic skill, applying broad training which is inefficient and unable to target specific reasoning errors. We introduce Reasoning Editing, a paradigm for selectively modifying specific reasoning patterns in LLMs while preserving other reasoning pathways. This task presents a fundamental trade-off between Generality, the ability of an edit to generalize across different tasks sharing the same reasoning pattern, and Locality, the ability to preserve other reasoning capabilities. Through systematic investigation, we uncover the Circuit-Interference Law: Edit interference between reasoning patterns is proportional to the overlap of their neural circuits. Guided by this principle, we propose REdit, the first framework to actively reshape neural circuits before editing, thereby modulating interference between reasoning patterns and mitigating the trade-off. REdit integrates three components: (i) Contrastive Circuit Reshaping, which directly addresses the generality-locality trade-off by disentangling overlapping circuits; (ii) Meta-Contrastive Learning, which extends transferability to novel reasoning patterns; and (iii) Dual-Level Protection, which preserves preexisting abilities by constraining reshaping update directions and regularizing task-level predictions. Extensive experiments with Qwen-2.5-3B on propositional logic reasoning tasks across three difficulty levels demonstrate that REdit consistently achieves superior generality and locality compared to baselines, with additional validation in mathematics showing broader potential. Our code is available at https://github.com/LzyFischer/REdit.

cs.CL↗

arXiv2025

Tropical geometry in torus bundles

We formulate and prove an analogue of the balancing condition for the tropicalization of a curve in a torus bundle $X$. We find that the usual balancing condition fails when the bundle is non-trivial and that the failure is captured by the first Chern classes of the line bundles associated to $X$. We discuss a geometric perspective of tropicalization where the weights arise naturally from intersection theory. The relations between divisors on a toric variety bundle put constraints on these weights which leads to the balancing condition.

math.AG↗

arXiv2025

Unifying and Optimizing Data Values for Selection via Sequential Decision-Making

Data selection has emerged as a crucial downstream application of data valuation, yet the theoretical foundations for using data values in selection remain underexplored. We reformulate data selection as a sequential decision-making problem where the optimal selection sequence arises from dynamic programming, and data values can be understood as encodings of this optimal sequence. This framework unifies and reinterprets existing methods like Data Shapley through the lens of approximate dynamic programming, revealing them as myopic linear approximations to the sequential problem. We further analyze how selection optimality degrades with utility curvature under submodularity, explaining when and why these approximations fail. To bridge theory and practice, we propose an efficient bipartite graph-based surrogate that preserves submodular structure while enabling scalable greedy selection with provable guarantees. Experiments on classical ML benchmarks and large-scale LLM fine-tuning data selection demonstrate substantial improvements over existing methods. Code is publicly available at https://github.com/frankhlchi/SeqDataVal

cs.AI↗

arXiv2023

A Clustering Algorithm to Organize Satellite Hotspot Data for the Purpose of Tracking Bushfires Remotely

This paper proposes a spatiotemporal clustering algorithm and its implementation in the R package spotoroo. This work is motivated by the catastrophic bushfires in Australia throughout the summer of 2019-2020 and made possible by the availability of satellite hotspot data. The algorithm is inspired by two existing spatiotemporal clustering algorithms but makes enhancements to cluster points spatially in conjunction with their movement across consecutive time periods. It also allows for the adjustment of key parameters, if required, for different locations and satellite data sources. Bushfire data from Victoria, Australia, is used to illustrate the algorithm and its use within the package.

cs.LG↗

arXiv2021

Towards Integrating Fairness Transparently in Industrial Applications

Numerous Machine Learning (ML) bias-related failures in recent years have led to scrutiny of how companies incorporate aspects of transparency and accountability in their ML lifecycles. Companies have a responsibility to monitor ML processes for bias and mitigate any bias detected, ensure business product integrity, preserve customer loyalty, and protect brand image. Challenges specific to industry ML projects can be broadly categorized into principled documentation, human oversight, and need for mechanisms that enable information reuse and improve cost efficiency. We highlight specific roadblocks and propose conceptual solutions on a per-category basis for ML practitioners and organizational subject matter experts. Our systematic approach tackles these challenges by integrating mechanized and human-in-the-loop components in bias detection, mitigation, and documentation of projects at various stages of the ML lifecycle. To motivate the implementation of our system -- SIFT (System to Integrate Fairness Transparently) -- we present its structural primitives with an example real-world use case on how it can be used to identify potential biases and determine appropriate mitigation strategies in a participatory manner.

cs.CY↗

arXiv2020

SOAR: Simultaneous Or of And Rules for Classification of Positive & Negative Classes

Algorithmic decision making has proliferated and now impacts our daily lives in both mundane and consequential ways. Machine learning practitioners make use of a myriad of algorithms for predictive models in applications as diverse as movie recommendations, medical diagnoses, and parole recommendations without delving into the reasons driving specific predictive decisions. Machine learning algorithms in such applications are often chosen for their superior performance, however popular choices such as random forest and deep neural networks fail to provide an interpretable understanding of the predictive model. In recent years, rule-based algorithms have been used to address this issue. Wang et al. (2017) presented an or-of-and (disjunctive normal form) based classification technique that allows for classification rule mining of a single class in a binary classification; this method is also shown to perform comparably to other modern algorithms. In this work, we extend this idea to provide classification rules for both classes simultaneously. That is, we provide a distinct set of rules for both positive and negative classes. In describing this approach, we also present a novel and complete taxonomy of classifications that clearly capture and quantify the inherent ambiguity in noisy binary classifications in the real world. We show that this approach leads to a more granular formulation of the likelihood model and a simulated-annealing based optimization achieves classification performance competitive with comparable techniques. We apply our method to synthetic as well as real world data sets to compare with other related methods that demonstrate the utility of our proposal.

stat.ML↗