Daehee Han
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
Daehee Han has a total of 37 co-authors. Here are the top co-authors by number of shared publications:
- Chanyeol Choi
LinqAlpha, New York, USA
2 shared publications - Junhyeong Lee
UNIST, Ulsan, Republic of Korea
2 shared publications - Sangjin Jin
UNIST, Ulsan, Republic of Korea
2 shared publications - Yongjae Lee
UNIST, Ulsan, Republic of Korea
2 shared publications - Adam M. Leventhal
Preventive Medicine, Keck School of Medicine of the University of Southern California, Los Angeles, California, USA
1 shared publication - Alyssa F. Harlow
Department of Population and Public Health Sciences, Keck School of Medicine, University of Southern California, Los Angeles, California, USA
1 shared publication
Latest Publications
- Open Access
- 13 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 973-981
From Hindsight to Foresight: Adapting Reusable Reasoning Structures from Recurring Financial Events
- 13 November 2026
- Published by Association for Computing Machinery (ACM)
- p. 973-981
Open AccessFrom Hindsight to Foresight: Adapting Reusable Reasoning Structures from Recurring Financial Events
- Open Access
- 19 September 2026
- Published by Elsevier in SSRN Electronic Journal
From Hindsight to Foresight: Adapting Reusable Reasoning Structures from Recurring Financial Events
Forecasting real-world events requires reasoning under uncertainty from information available at a fixed cutoff. Recent work has improved how large language models (LLMs) retrieve evidence and produce probabilistic forecasts, but each question is treated independently even when similar events recur. Resolved occurrences carry experience for later forecasts, yet past outcomes and case summaries do not preserve the relationships that made an earlier occurrence informative. We develop ReFoRM (Recurrent-event Forecasting with Reasoning Memory), which organizes reasoning from resolved events into predictive paths, retrieves relevant paths for a later occurrence, and reconstructs their relations using evidence at the new forecast cutoff. We evaluate ReFoRM on the Recurring Financial Events Set, which contains 300 questions across 40 recurring financial event families and uses earlier occurrences as hindsight for 100 forecast targets. Across five LLMs and six baselines, ReFoRM improves categorical and probabilistic forecasting performance. Further analyses show that validating historical relations with current evidence and reconstructing them enhances forecasting. Reasoning evaluation also demonstrates stronger evidence grounding, logical validity, and mechanism specificity.
- 19 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessFrom Hindsight to Foresight: Adapting Reusable Reasoning Structures from Recurring Financial Events
Forecasting real-world events requires reasoning under uncertainty from information available at a fixed cutoff. Recent work has improved how large language models (LLMs) retrieve evidence and produce probabilistic forecasts, but each question is treated independently even when similar events recur. Resolved occurrences carry experience for later forecasts, yet past outcomes and case summaries do not preserve the relationships that made an earlier occurrence informative. We develop ReFoRM (Recurrent-event Forecasting with Reasoning Memory), which organizes reasoning from resolved events into predictive paths, retrieves relevant paths for a later occurrence, and reconstructs their relations using evidence at the new forecast cutoff. We evaluate ReFoRM on the Recurring Financial Events Set, which contains 300 questions across 40 recurring financial event families and uses earlier occurrences as hindsight for 100 forecast targets. Across five LLMs and six baselines, ReFoRM improves categorical and probabilistic forecasting performance. Further analyses show that validating historical relations with current evidence and reconstructing them enhances forecasting. Reasoning evaluation also demonstrates stronger evidence grounding, logical validity, and mechanism specificity.
- Open Access
- 10 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 1449-1457
Amortized Baseline Selection via Rank-Revealing QR for Efficient Model Explanation
- 10 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 1449-1457
Open AccessAmortized Baseline Selection via Rank-Revealing QR for Efficient Model Explanation
- 25 November 2024
- Published by Elsevier in Applied Soft Computing
- Vol. 169
Normalizing flow-based latent space mapping for implicit pattern authentication on mobile devices
- 25 November 2024
- Published by Elsevier in Applied Soft Computing
- Vol. 169
Normalizing flow-based latent space mapping for implicit pattern authentication on mobile devices
- 16 July 2024
- Published by Oxford University Press (OUP) in American Journal of Epidemiology
- Vol. 194 (3) , 746-754
Vaping transitions and incident depressive symptoms among young adults: a marginal structural model analysis
The extent to which vaping influences depression is unclear, but could be estimated through application of novel epidemiologic methods. Among a prospective cohort of young adults from California who screened negative for depression, we estimated repeated measures marginal structural models to examine the association of four vaping transitions from time T to T+1 (persistent use, discontinuation, initiation, persistent nonuse) with risk of clinically significant depressive symptoms at T+1, simultaneously across three ~1.5 year time-intervals between 2017-2021. Stabilized inverse probability of treatment and censoring weights adjusted for time-dependent confounders and selection bias. Among n=3,496 observations (1,806 participants, mean pooled baseline age=19.5), 8.1% reported persistent vaping from T to T+1, 6.2% reported discontinuation (i.e., use at T and no use at T+1), 6.5% initiated e-cigarettes (i.e., no use at T and use at T+1), and 79.2% reported persistent nonuse at both time-points. Compared to persistent vaping at two waves, persistent nonuse (RR=0.76, 95%CI:0.62-0.93) and discontinuation (RR=0.71, 95%CI:0.52-0.96) were associated with lower risk of depression. Associations were robust to sensitivity analyses, including restricting to tobacco naïve participants and varying temporal assumptions to reduce potential for reverse causation. Young adults who consistently avoid or discontinue vaping may be protected from depressive symptom occurrence.
- 16 July 2024
- Published by Oxford University Press (OUP) in American Journal of Epidemiology
- Vol. 194 (3) , 746-754
Vaping transitions and incident depressive symptoms among young adults: a marginal structural model analysis
The extent to which vaping influences depression is unclear, but could be estimated through application of novel epidemiologic methods. Among a prospective cohort of young adults from California who screened negative for depression, we estimated repeated measures marginal structural models to examine the association of four vaping transitions from time T to T+1 (persistent use, discontinuation, initiation, persistent nonuse) with risk of clinically significant depressive symptoms at T+1, simultaneously across three ~1.5 year time-intervals between 2017-2021. Stabilized inverse probability of treatment and censoring weights adjusted for time-dependent confounders and selection bias. Among n=3,496 observations (1,806 participants, mean pooled baseline age=19.5), 8.1% reported persistent vaping from T to T+1, 6.2% reported discontinuation (i.e., use at T and no use at T+1), 6.5% initiated e-cigarettes (i.e., no use at T and use at T+1), and 79.2% reported persistent nonuse at both time-points. Compared to persistent vaping at two waves, persistent nonuse (RR=0.76, 95%CI:0.62-0.93) and discontinuation (RR=0.71, 95%CI:0.52-0.96) were associated with lower risk of depression. Associations were robust to sensitivity analyses, including restricting to tobacco naïve participants and varying temporal assumptions to reduce potential for reverse causation. Young adults who consistently avoid or discontinue vaping may be protected from depressive symptom occurrence.