Yongjae Lee

ORCID: 0000-0002-5411-4340
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Financial Markets and Investment Strategies
  • Risk and Portfolio Optimization
  • Microbial Natural Products and Biosynthesis
  • Stock Market Forecasting Methods
  • Genomics and Phylogenetic Studies
  • RNA and protein synthesis mechanisms
  • Reservoir Engineering and Simulation Methods
  • Complex Systems and Time Series Analysis
  • Cybercrime and Law Enforcement Studies
  • Housing Market and Economics
  • Economic theories and models
  • Time Series Analysis and Forecasting
  • Spam and Phishing Detection
  • Stochastic processes and financial applications
  • Recommender Systems and Techniques
  • Network Security and Intrusion Detection
  • Advanced Bandit Algorithms Research
  • Educational Systems and Policies
  • Technology and Data Analysis
  • Authorship Attribution and Profiling
  • Insurance and Financial Risk Management
  • Topic Modeling
  • Interactive and Immersive Displays
  • Financial Risk and Volatility Modeling
  • Financial Literacy, Pension, Retirement Analysis

Arizona State University
2023-2025

Ulsan National Institute of Science and Technology
2014-2024

Korea Advanced Institute of Science and Technology
2013-2024

Samsung (United States)
2020-2024

University of California, Davis
2022

Korea University
2017-2020

Ulsan College
2020

Seoul National University
2018

Anyang University
2014

Hanyang University
2014

Compared to the traditional machine learning models, deep neural networks (DNN) are known be highly sensitive choice of hyperparameters. While required time and effort for manual tuning has been rapidly decreasing well developed commonly used DNN architectures, undoubtedly hyperparameter optimization will continue a major burden whenever new architecture needs designed, task solved, dataset addressed, or an existing improved further. For general problems, numerous automated solutions have...

10.1109/access.2020.2981072 article EN cc-by IEEE Access 2020-01-01

Abstract Streptomyces are Gram-positive bacteria of significant industrial importance due to their ability produce a wide range antibiotics and bioactive secondary metabolites. Recent advances in genome mining have revealed that genomes possess large number unexplored silent metabolite biosynthetic gene clusters (smBGCs). This indicates continue be an invaluable source for new drug discovery. Here, we present high-quality sequences 22 species eight different venezuelae strains assembled by...

10.1038/s41597-020-0395-9 article EN cc-by Scientific Data 2020-02-13

Microbial coculture to mimic the ecological habitat has been suggested as an approach elucidate effect of microbial interaction on secondary metabolite biosynthesis Streptomyces. However, because chemical complexity during coculture, underlying mechanisms are largely unknown. Here, we found that iron competition triggered antibiotic in Streptomyces coelicolor with Myxococcus xanthus. During M. xanthus enhanced production a siderophore, myxochelin, leading dominate scavenging and S....

10.1038/s41396-020-0594-6 article EN cc-by The ISME Journal 2020-01-28

Recent advances in neuromorphic computing have established a computational framework that removes the processor-memory bottleneck evident traditional von Neumann computing. Moreover, contemporary photonic circuits addressed limitations of electrical platforms to offer energy-efficient and parallel interconnects independently distance. When employed as synaptic with reconfigurable elements, they can an analog platform capable arbitrary linear matrix operations, including multiply–accumulate...

10.1063/5.0072090 article EN cc-by APL Photonics 2022-02-07

Recent research has suggested that there are clear differences in the language used Dark Web compared to of Surface Web. As studies on commonly require textual analysis domain, models specific may provide valuable insights researchers. In this work, we introduce DarkBERT, a model pretrained data. We describe steps taken filter and compile text data train DarkBERT combat extreme lexical structural diversity be detrimental building proper representation domain. evaluate its vanilla counterpart...

10.18653/v1/2023.acl-long.415 article EN cc-by 2023-01-01

Abstract Determining transcriptional and translational regulatory elements in GC-rich Streptomyces genomes is essential to elucidating the complex networks that govern secondary metabolite biosynthetic gene cluster (BGC) expression. However, information about such has been limited for genomes. To address this limitation, a high-quality genome sequence of β-lactam antibiotic-producing clavuligerus ATCC 27 064 completed, which contains 7163 newly annotated genes. This provides fundamental...

10.1093/nar/gkz471 article EN cc-by-nc Nucleic Acids Research 2019-05-17

This paper addresses the critical disconnect between prediction and decision quality in portfolio optimization by integrating Large Language Models (LLMs) with decision-focused learning. We demonstrate both theoretically empirically that minimizing error alone leads to suboptimal decisions. aim exploit representational power of LLMs for investment An attention mechanism processes asset relationships, temporal dependencies, macro variables, which are then directly integrated into a layer....

10.48550/arxiv.2502.00828 preprint EN arXiv (Cornell University) 2025-02-02

Phishing often targets victims through visually perturbed texts to bypass security systems. The noise contained in these functions as an adversarial attack, designed deceive language models and hinder their ability accurately interpret the content. However, since it is difficult obtain sufficient phishing cases, previous studies have used synthetic datasets that do not contain real-world cases. In this study, we propose BitAbuse dataset, which includes address limitations of research. Our...

10.48550/arxiv.2502.05225 preprint EN arXiv (Cornell University) 2025-02-06

In practice, including large number of assets in mean-variance portfolios can lead to higher transaction costs and management fees. To address this, one common approach is select a smaller subset from the larger pool, constructing more efficient portfolios. As solution, we propose new asset selection heuristic which generates pre-defined list candidates using surrogate formulation re-optimizes cardinality-constrained tangent portfolio with these selected assets. This method enables faster...

10.48550/arxiv.2502.11701 preprint EN arXiv (Cornell University) 2025-02-17

10.1109/wacv61041.2025.00266 article EN 2022 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2025-02-26

Tabular data poses unique challenges due to its heterogeneous nature, combining both continuous and categorical variables. Existing approaches often struggle effectively capture the underlying structure relationships within such data. We propose GFTab (Geodesic Flow Kernels for Semi-Supervised Learning on Mixed-Variable Dataset), a semi-supervised framework specifically designed tabular datasets. incorporates three key innovations: 1) Variable-specific corruption methods tailored distinct...

10.1609/aaai.v39i17.33928 article EN Proceedings of the AAAI Conference on Artificial Intelligence 2025-04-11

Abstract Ocular adnexal lymphoma (OAL) is a mostly extranodal marginal zone (EMZL). Recent findings have suggested an association between Chlamydia psittaci (Cp ) infection and OAL. We sought to confirm this issue analyze the clinicopathologic characteristics of OAL in Korea. Between 1993 2004, 33 cases were identified at Asan Medical Center, Seoul, DNA was extracted from paraffin‐embedded tissues, touchdown enzyme time release polymerase chain reaction performed identify three species ( Cp,...

10.1002/ajh.20962 article EN American Journal of Hematology 2007-06-14

RNA sequencing techniques have enabled the systematic elucidation of gene expression (RNA-Seq), transcription start sites (differential RNA-Seq), transcript 3′ ends (Term-Seq), and post-transcriptional processes (ribosome profiling). The main challenge transcriptomic studies is to remove ribosomal RNAs (rRNAs), which comprise more than 90% total in a cell. Here, we report low-cost robust bacterial rRNA depletion method, RiboRid, based on enzymatic degradation by thermostable RNase H. This...

10.1371/journal.pgen.1009821 article EN cc-by PLoS Genetics 2021-09-27

Streptomyces lividans is an attractive host for production of heterologous proteins and secondary metabolites other species. To fully harness the industrial potential S. lividans, understanding its metabolism genetic regulatory elements essential. This study aimed to determine transcription unit (TU) architecture elucidate diverse elements, including promoters, ribosome binding sites, 5'-untranslated regions, terminators. Total 1,978 start sites 1,640 transcript 3'-end positions were...

10.3389/fmicb.2019.02074 article EN cc-by Frontiers in Microbiology 2019-09-06

The mean–variance model is widely acknowledged as the foundation of portfolio allocation because it provides a framework for analyzing trade-off between risk and return gaining diversification benefits. Despite well-known shortcomings model, often starting point making asset decisions. In this article, authors briefly review optimization approaches resolving its limitations by demonstrating backtest results on allocation. Feedback from managers also included to explain how methods are...

10.3905/jpm.2021.1.219 article EN The Journal of Portfolio Management 2021-02-12

Machine learning has been widely used in the asset management industry to improve operations and make data-driven decisions. This article provides an overview of machine for by presenting various models context their applications, including general classification regression, time series forecasting, natural language processing, dimension reduction, reinforcement learning, data generation, recommendation, clustering. Additionally, it highlights challenges implementing management, such as...

10.3905/jpm.2023.1.526 article EN The Journal of Portfolio Management 2023-07-30

In this paper, we propose a goal-based investment model that is suitable for personalized wealth management. The only requires few intuitive inputs such as size of wealth, amount, and consumption goals from individual investors. particular, priority level can be assigned to each goal the provides holistic solution based on sequential approach starting with highest priority. This allows strict prioritization by maximizing probability achieving higher are not affected lower priorities....

10.1080/14697688.2019.1662079 article EN Quantitative Finance 2019-10-31

Streptomyces are efficient producers of various bioactive compounds, which mostly synthesized by their secondary metabolite biosynthetic gene clusters (smBGCs). The smBGCs tightly controlled complex regulatory systems at transcriptional and translational levels to effectively utilize precursors that supplied primary metabolism. Thus, dynamic changes in expression response cellular status both the should be elucidated directly reflect protein levels, rapid downstream responses, energy costs....

10.1038/s41597-020-0476-9 article EN cc-by Scientific Data 2020-05-08

As Non-Fungible Tokens (NFTs) continue to grow in popularity, NFT users have become targets of phishing scammers, called drainers.Over the last year, $100 million worth NFTs were stolen by drainers, and their presence remains a serious threat trading space.However, no work has yet comprehensively investigated behaviors drainers ecosystem.In this paper, we present first study on behavior introduce dedicated drainer detection system.We collect 127M transaction data from Ethereum blockchain...

10.14722/ndss.2024.24888 article EN 2024-01-01
Coming Soon ...