Frank Rudzicz
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Frank Rudzicz has a total of 229 co-authors. Here are the top co-authors by number of shared publications:
- Zining Zhu
Department of Computer Science University of Toronto
6 shared publications - Jekaterina Novikova
Winterlight Labs Toronto Canada
5 shared publications - Robert E. Mercer
The University of Western Ontario
5 shared publications - Puneet Seth
Department of Family Medicine, McMaster University, Hamilton, ON, Canada
4 shared publications - Aparna Balagopalan
Winterlight Labs Toronto Canada
3 shared publications - Mohamed Abdalla
Department of Medicine, Faculty of Medicine & Dentistry, University of Alberta, 8-130 Clinical Sciences Building, 11350 83 Ave NW, Edmonton, AB, T6G2G3, Canada, 1 780-492-7002, 1 780-492-6088
3 shared publications
Latest Publications
- Open Access
- 28 October 2026
- Published by Elsevier in JACC: Advances
- Vol. 5 (10)
Incidence of Dementia and Subsequent Outcomes in Patients With Heart Failure
- 28 October 2026
- Published by Elsevier in JACC: Advances
- Vol. 5 (10)
Open AccessIncidence of Dementia and Subsequent Outcomes in Patients With Heart Failure
- Open Access
- 8 October 2026
- Published by Elsevier in Intelligent Systems with Applications
- Vol. 32
Inside summary evaluation: Understanding human and GPT judgments to improve LLM-based evaluation
- 8 October 2026
- Published by Elsevier in Intelligent Systems with Applications
- Vol. 32
Open AccessInside summary evaluation: Understanding human and GPT judgments to improve LLM-based evaluation
- Open Access
- 14 May 2026
- Published by Elsevier in SSRN Electronic Journal
LUG: Low-frequency Upscaling Guidance forPlug-and-Play Enhancement of Flow and DiffusionTransformers
A significant quality gap persists between unconditional generation and classifier-freeguidance (CFG) sampling in transformer-based diffusion and flow-matching models.Through frequency-domain analysis, we show that CFG sampling establishes a coher-ent low-frequency foundation corresponding to global shapes and layouts earlier andmore effectively than unconditional sampling. This strong foundation then supportssuperior refinement of high-frequency details. Inspired by this insight, we introduceLow-frequency Upscaling Guidance (LUG), a training-free method that amplifies low-frequency components in the model’s internal representations. LUG applies a simple,constant scaling factor to the attention outputs of pretrained Diffusion Transformer(DiT) and Scalable Interpolant Transformer (SiT) blocks, which our analysis identifiesas the primary carriers of low-frequency information. This zero-cost, plug-and-playmodification consistently improves sample quality across DiT and SiT models on Im-ageNet at 256 × 256 and 512 × 512 resolutions, improving Fréchet Inception Distance(FID) by up to 50% for unconditional generation. In addition, LUG also improvesCFG sampling in terms of Inception score and Precision with negligible or setting-dependent recall trade-offs, establishing it as a robust and efficient enhancement fortransformer-based diffusion and flow models.
- 14 May 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessLUG: Low-frequency Upscaling Guidance forPlug-and-Play Enhancement of Flow and DiffusionTransformers
A significant quality gap persists between unconditional generation and classifier-freeguidance (CFG) sampling in transformer-based diffusion and flow-matching models.Through frequency-domain analysis, we show that CFG sampling establishes a coher-ent low-frequency foundation corresponding to global shapes and layouts earlier andmore effectively than unconditional sampling. This strong foundation then supportssuperior refinement of high-frequency details. Inspired by this insight, we introduceLow-frequency Upscaling Guidance (LUG), a training-free method that amplifies low-frequency components in the model’s internal representations. LUG applies a simple,constant scaling factor to the attention outputs of pretrained Diffusion Transformer(DiT) and Scalable Interpolant Transformer (SiT) blocks, which our analysis identifiesas the primary carriers of low-frequency information. This zero-cost, plug-and-playmodification consistently improves sample quality across DiT and SiT models on Im-ageNet at 256 × 256 and 512 × 512 resolutions, improving Fréchet Inception Distance(FID) by up to 50% for unconditional generation. In addition, LUG also improvesCFG sampling in terms of Inception score and Precision with negligible or setting-dependent recall trade-offs, establishing it as a robust and efficient enhancement fortransformer-based diffusion and flow models.
- Open Access
- 3 March 2026
- Published by Elsevier in SSRN Electronic Journal
Limits of Convergence-Rate Control for Open-Weight Safety
Open-weight foundation models can be fine-tuned for harmful purposes after release, yet no existing training resistance methods provide theoretical guarantees. Treating these interventions as convergence-rate control problems allows us to connect optimization speed to the spectral structure of model weights. We leverage this insight to develop a novel understanding of convergence rate control through spectral reparameterization and derive an algorithm, SpecDef, that can both provably and empirically slow first-and secondorder optimization in non-adversarial settings. In adversarial settings, we establish a fundamental limit on a broad class of convergence rate control methods including our own: an attacker with sufficient knowledge can restore fast convergence at a linear increase in model size. In order to overcome this limitation, future works will need to investigate methods that are not equivalent to controlling convergence rate.
- 3 March 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessLimits of Convergence-Rate Control for Open-Weight Safety
Open-weight foundation models can be fine-tuned for harmful purposes after release, yet no existing training resistance methods provide theoretical guarantees. Treating these interventions as convergence-rate control problems allows us to connect optimization speed to the spectral structure of model weights. We leverage this insight to develop a novel understanding of convergence rate control through spectral reparameterization and derive an algorithm, SpecDef, that can both provably and empirically slow first-and secondorder optimization in non-adversarial settings. In adversarial settings, we establish a fundamental limit on a broad class of convergence rate control methods including our own: an attacker with sufficient knowledge can restore fast convergence at a linear increase in model size. In order to overcome this limitation, future works will need to investigate methods that are not equivalent to controlling convergence rate.
- Open Access
MedSynth: Realistic, Synthetic Medical Dialogue-Note Pairs
Physicians spend significant time documenting clinical encounters, a burden that contributes to professional burnout. To address this, robust automation tools for medical documentation are crucial. We introduce MedSynth -- a novel dataset of synthetic medical dialogues and notes designed to advance the Dialogue-to-Note (Dial-2-Note) and Note-to-Dialogue (Note-2-Dial) tasks. Informed by an extensive analysis of disease distributions, this dataset includes over 10,000 dialogue-note pairs covering over 2000 ICD-10 codes. We demonstrate that our dataset markedly enhances the performance of models in generating medical notes from dialogues, and dialogues from medical notes. The dataset provides a valuable resource in a field where open-access, privacy-compliant, and diverse training data are scarce. Code is available at https://github.com/ahmadrezarm/MedSynth/tree/main and the dataset is available at https://huggingface.co/datasets/Ahmad0067/MedSynth.
Open AccessMedSynth: Realistic, Synthetic Medical Dialogue-Note Pairs
Physicians spend significant time documenting clinical encounters, a burden that contributes to professional burnout. To address this, robust automation tools for medical documentation are crucial. We introduce MedSynth -- a novel dataset of synthetic medical dialogues and notes designed to advance the Dialogue-to-Note (Dial-2-Note) and Note-to-Dialogue (Note-2-Dial) tasks. Informed by an extensive analysis of disease distributions, this dataset includes over 10,000 dialogue-note pairs covering over 2000 ICD-10 codes. We demonstrate that our dataset markedly enhances the performance of models in generating medical notes from dialogues, and dialogues from medical notes. The dataset provides a valuable resource in a field where open-access, privacy-compliant, and diverse training data are scarce. Code is available at https://github.com/ahmadrezarm/MedSynth/tree/main and the dataset is available at https://huggingface.co/datasets/Ahmad0067/MedSynth.
- Open Access
Contrastive Similarity Learning for Market Forecasting: The ContraSim Framework
We introduce the Contrastive Similarity Space Embedding Algorithm (ContraSim), a novel framework for uncovering the global semantic relationships between daily financial headlines and market movements. ContraSim operates in two key stages: (I) Weighted Headline Augmentation, which generates augmented financial headlines along with a semantic fine-grained similarity score, and (II) Weighted Self-Supervised Contrastive Learning (WSSCL), an extended version of classical self-supervised contrastive learning that uses the similarity metric to create a refined weighted embedding space. This embedding space clusters semantically similar headlines together, facilitating deeper market insights. Empirical results demonstrate that integrating ContraSim features into financial forecasting tasks improves classification accuracy from WSJ headlines by 7%. Moreover, leveraging an information density analysis, we find that the similarity spaces constructed by ContraSim intrinsically cluster days with homogeneous market movement directions, indicating that ContraSim captures market dynamics independent of ground truth labels. Additionally, ContraSim enables the identification of historical news days that closely resemble the headlines of the current day, providing analysts with actionable insights to predict market trends by referencing analogous past events.
Open AccessContrastive Similarity Learning for Market Forecasting: The ContraSim Framework
We introduce the Contrastive Similarity Space Embedding Algorithm (ContraSim), a novel framework for uncovering the global semantic relationships between daily financial headlines and market movements. ContraSim operates in two key stages: (I) Weighted Headline Augmentation, which generates augmented financial headlines along with a semantic fine-grained similarity score, and (II) Weighted Self-Supervised Contrastive Learning (WSSCL), an extended version of classical self-supervised contrastive learning that uses the similarity metric to create a refined weighted embedding space. This embedding space clusters semantically similar headlines together, facilitating deeper market insights. Empirical results demonstrate that integrating ContraSim features into financial forecasting tasks improves classification accuracy from WSJ headlines by 7%. Moreover, leveraging an information density analysis, we find that the similarity spaces constructed by ContraSim intrinsically cluster days with homogeneous market movement directions, indicating that ContraSim captures market dynamics independent of ground truth labels. Additionally, ContraSim enables the identification of historical news days that closely resemble the headlines of the current day, providing analysts with actionable insights to predict market trends by referencing analogous past events.
- Open Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 27477-27490
Trustworthy Medical Question Answering: An Evaluation-Centric Survey
Yinuo Wang, Baiyang Wang, Robert Mercer, Frank Rudzicz, Sudipta Singha Roy, Pengjie Ren, Zhumin Chen, Xindi Wang. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 27477-27490
Open AccessTrustworthy Medical Question Answering: An Evaluation-Centric Survey
Yinuo Wang, Baiyang Wang, Robert Mercer, Frank Rudzicz, Sudipta Singha Roy, Pengjie Ren, Zhumin Chen, Xindi Wang. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.
- Open Access
- 23 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 20374-20412
MLHOps: Machine Learning Health Operations
Machine Learning Health Operations (MLHOps) is the combination of processes for reliable, efficient, usable, and ethical deployment and maintenance of machine learning models in healthcare settings. This paper provides both a survey of work in this area and guidelines for developers and clinicians to deploy and maintain their own models in clinical practice. We cover the foundational concepts of general machine learning operations and describe the initial setup of MLHOps pipelines (including data sources, preparation, engineering, and tools). We then describe long-term monitoring and updating (including data distribution shifts and model updating) and ethical considerations (including bias, fairness, interpretability, and privacy). This work therefore provides guidance across the full pipeline of MLHOps from conception to initial and ongoing deployment. We also include a checklist to ensure thorough verification of each step in the process.
- 23 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 20374-20412
Open AccessMLHOps: Machine Learning Health Operations
Machine Learning Health Operations (MLHOps) is the combination of processes for reliable, efficient, usable, and ethical deployment and maintenance of machine learning models in healthcare settings. This paper provides both a survey of work in this area and guidelines for developers and clinicians to deploy and maintain their own models in clinical practice. We cover the foundational concepts of general machine learning operations and describe the initial setup of MLHOps pipelines (including data sources, preparation, engineering, and tools). We then describe long-term monitoring and updating (including data distribution shifts and model updating) and ethical considerations (including bias, fairness, interpretability, and privacy). This work therefore provides guidance across the full pipeline of MLHOps from conception to initial and ongoing deployment. We also include a checklist to ensure thorough verification of each step in the process.
- Open Access
Show, Don't Tell: Uncovering Implicit Character Portrayal using LLMs
Tools for analyzing character portrayal in fiction are valuable for writers and literary scholars in developing and interpreting compelling stories. Existing tools, such as visualization tools for analyzing fictional characters, primarily rely on explicit textual indicators of character attributes. However, portrayal is often implicit, revealed through actions and behaviors rather than explicit statements. We address this gap by leveraging large language models (LLMs) to uncover implicit character portrayals. We start by generating a dataset for this task with greater cross-topic similarity, lexical diversity, and narrative lengths than existing narrative text corpora such as TinyStories and WritingPrompts. We then introduce LIIPA (LLMs for Inferring Implicit Portrayal for Character Analysis), a framework for prompting LLMs to uncover character portrayals. LIIPA can be configured to use various types of intermediate computation (character attribute word lists, chain-of-thought) to infer how fictional characters are portrayed in the source text. We find that LIIPA outperforms existing approaches, and is more robust to increasing character counts (number of unique persons depicted) due to its ability to utilize full narrative context. Lastly, we investigate the sensitivity of portrayal estimates to character demographics, identifying a fairness-accuracy tradeoff among methods in our LIIPA framework -- a phenomenon familiar within the algorithmic fairness literature. Despite this tradeoff, all LIIPA variants consistently outperform non-LLM baselines in both fairness and accuracy. Our work demonstrates the potential benefits of using LLMs to analyze complex characters and to better understand how implicit portrayal biases may manifest in narrative texts.
Open AccessShow, Don't Tell: Uncovering Implicit Character Portrayal using LLMs
Tools for analyzing character portrayal in fiction are valuable for writers and literary scholars in developing and interpreting compelling stories. Existing tools, such as visualization tools for analyzing fictional characters, primarily rely on explicit textual indicators of character attributes. However, portrayal is often implicit, revealed through actions and behaviors rather than explicit statements. We address this gap by leveraging large language models (LLMs) to uncover implicit character portrayals. We start by generating a dataset for this task with greater cross-topic similarity, lexical diversity, and narrative lengths than existing narrative text corpora such as TinyStories and WritingPrompts. We then introduce LIIPA (LLMs for Inferring Implicit Portrayal for Character Analysis), a framework for prompting LLMs to uncover character portrayals. LIIPA can be configured to use various types of intermediate computation (character attribute word lists, chain-of-thought) to infer how fictional characters are portrayed in the source text. We find that LIIPA outperforms existing approaches, and is more robust to increasing character counts (number of unique persons depicted) due to its ability to utilize full narrative context. Lastly, we investigate the sensitivity of portrayal estimates to character demographics, identifying a fairness-accuracy tradeoff among methods in our LIIPA framework -- a phenomenon familiar within the algorithmic fairness literature. Despite this tradeoff, all LIIPA variants consistently outperform non-LLM baselines in both fairness and accuracy. Our work demonstrates the potential benefits of using LLMs to analyze complex characters and to better understand how implicit portrayal biases may manifest in narrative texts.