T-Robinson Spaces: Structure, Recognition, and Applications to Real Data
研究T-Robinson空间，通过图和超图理论性质对其表征，并开发了识别算法以处理现实数据中的层次结构。
0 citationsRead paper
Selected work
Representative Papers
Querying an astronomical database using large language models: the ALeRCE text-to-SQL system
This work addresses the high barrier posed by SQL-based querying of astronomical databases for non-expert users. The authors propose a large language model (LLM)-driven text-to-SQL system enabling natural language queries over the ALeRCE database. Their approach employs a four-module, stepwise generation framework—comprising schema linking, query classification, prompt decomposition, and self-correction—which substantially outperforms end-to-end baselines and significantly reduces execution errors. Evaluation on 110 annotated samples across 13 LLMs demonstrates strong performance: on models such as Claude Opus 4.6, exact match rates for row and column identifiers reach 0.97 and 0.94, respectively, for simple queries, while maintaining robustness on complex queries.
0 citationsRead paper
Automated Palynological Analysis System: Integrating Deep Metric Learning and $U^{2}$-Net Detection in $H\infty$ bright field microscopy
Traditional pollen analysis is time-consuming (4–6 hours per sample) and highly subjective. This study proposes an automated, high-throughput microscopic analysis system that integrates brightfield imaging, $H_\infty$ robust mechanical control, and a deep learning pipeline to enable efficient, accurate counting, classification, and morphological characterization of pollen from the Bio Bío region of Chile. The approach combines U²-Net for salient object detection with a DINOv2 vision transformer classifier based on deep metric learning, augmented by a gradient-weighted attention mechanism to generate interpretable texture and diagnostic features. The system achieves a classification recall of 95.8% and processes samples six times faster than human experts.
0 citationsRead paper
Talent is Everywhere, Mobility is Not: Mapping the Topological Anchors of Educational Pathways
This study investigates how socioeconomic background systematically constrains higher education mobility, using administrative data from 2.7 million Chilean students (2021–2024). Methodologically, it introduces a novel two-dimensional “educational space” topology integrating academic aptitude and familial socioeconomic status; applies unsupervised clustering to identify seven student archetypes; and combines dimensionality reduction, causal-informed path modeling, and large-scale administrative analytics. Results reveal pronounced structural geographic immobility among high-achieving, low-income students. The study quantifies, for the first time, the independent effects of family background on three distinct educational transitions: institutional enrollment choice, field-of-study selection, and interregional migration. It further develops a reusable, generalizable framework for global education equity policy evaluation. Findings have directly informed the design of multiple regional higher education equity interventions in Chile.
0 citationsRead paper
A streaming algorithm and hardware accelerator for top-K flow detection in network traffic
To address the challenge of efficiently and accurately identifying Top-K flows under skewed traffic distributions in high-speed networks, this paper proposes a streaming algorithm that integrates a dynamically updated TowerSketch with a parallel priority-queue array, implemented as a high-throughput hardware accelerator on an AMD Virtex-U280 FPGA. Our contributions are threefold: (1) an enhanced TowerSketch adapted to dynamic flow spectra, significantly improving estimation accuracy under skew; (2) a parallelizable priority-queue array design that alleviates on-chip memory bottlenecks; and (3) line-rate processing capability ≥200 Gbps—achieving one packet per cycle. Experimental evaluation on real-world traces demonstrates Top-K identification accuracy exceeding 0.94 and average relative error in frequency estimation below 1.96%.
0 citationsRead paper
Recent publications
Latest Papers
T-Robinson Spaces: Structure, Recognition, and Applications to Real Data
研究T-Robinson空间，通过图和超图理论性质对其表征，并开发了识别算法以处理现实数据中的层次结构。
0 citationsRead paper
Querying an astronomical database using large language models: the ALeRCE text-to-SQL system
This work addresses the high barrier posed by SQL-based querying of astronomical databases for non-expert users. The authors propose a large language model (LLM)-driven text-to-SQL system enabling natural language queries over the ALeRCE database. Their approach employs a four-module, stepwise generation framework—comprising schema linking, query classification, prompt decomposition, and self-correction—which substantially outperforms end-to-end baselines and significantly reduces execution errors. Evaluation on 110 annotated samples across 13 LLMs demonstrates strong performance: on models such as Claude Opus 4.6, exact match rates for row and column identifiers reach 0.97 and 0.94, respectively, for simple queries, while maintaining robustness on complex queries.
0 citationsRead paper
Automated Palynological Analysis System: Integrating Deep Metric Learning and $U^{2}$-Net Detection in $H\infty$ bright field microscopy
Traditional pollen analysis is time-consuming (4–6 hours per sample) and highly subjective. This study proposes an automated, high-throughput microscopic analysis system that integrates brightfield imaging, $H_\infty$ robust mechanical control, and a deep learning pipeline to enable efficient, accurate counting, classification, and morphological characterization of pollen from the Bio Bío region of Chile. The approach combines U²-Net for salient object detection with a DINOv2 vision transformer classifier based on deep metric learning, augmented by a gradient-weighted attention mechanism to generate interpretable texture and diagnostic features. The system achieves a classification recall of 95.8% and processes samples six times faster than human experts.
0 citationsRead paper
Talent is Everywhere, Mobility is Not: Mapping the Topological Anchors of Educational Pathways
This study investigates how socioeconomic background systematically constrains higher education mobility, using administrative data from 2.7 million Chilean students (2021–2024). Methodologically, it introduces a novel two-dimensional “educational space” topology integrating academic aptitude and familial socioeconomic status; applies unsupervised clustering to identify seven student archetypes; and combines dimensionality reduction, causal-informed path modeling, and large-scale administrative analytics. Results reveal pronounced structural geographic immobility among high-achieving, low-income students. The study quantifies, for the first time, the independent effects of family background on three distinct educational transitions: institutional enrollment choice, field-of-study selection, and interregional migration. It further develops a reusable, generalizable framework for global education equity policy evaluation. Findings have directly informed the design of multiple regional higher education equity interventions in Chile.
0 citationsRead paper
A streaming algorithm and hardware accelerator for top-K flow detection in network traffic
To address the challenge of efficiently and accurately identifying Top-K flows under skewed traffic distributions in high-speed networks, this paper proposes a streaming algorithm that integrates a dynamically updated TowerSketch with a parallel priority-queue array, implemented as a high-throughput hardware accelerator on an AMD Virtex-U280 FPGA. Our contributions are threefold: (1) an enhanced TowerSketch adapted to dynamic flow spectra, significantly improving estimation accuracy under skew; (2) a parallelizable priority-queue array design that alleviates on-chip memory bottlenecks; and (3) line-rate processing capability ≥200 Gbps—achieving one packet per cycle. Experimental evaluation on real-world traces demonstrates Top-K identification accuracy exceeding 0.94 and average relative error in frequency estimation below 1.96%.
0 citationsRead paper