FactorFlow: A Visual Analytics Workspace with Large Language Model-Assisted Interpretation for Factor Analysis
为了解决探索性因子分析中模型解释的主观性和复杂性问题，FactorFlow系统通过可视化仪表板和大语言模型辅助生成自然语言解释来帮助研究人员更有效地进行分析。
0 citationsRead paper
Selected work
Representative Papers
Cascading Effects of the COVID-19 Pandemic on Barangays in the Philippines
This study investigates the cascading impacts of the COVID-19 pandemic on Philippine barangays across multiple dimensions, including mobility, public services, economic and financial well-being, food security, education, and physical health. Drawing on survey data from 2,122 households collected between May and June 2021, the research pioneers an integrated approach combining Bayesian networks and influence diagrams, implemented in Python. Variables were subjected to feature clustering and z-score normalization to infer conditional dependencies and potential causal pathways among the multidimensional outcomes. The analysis identifies enhancing food production, stabilizing market prices, and expanding income-generating opportunities as the most effective interventions, underscoring the pivotal role of economic and food security policies. These findings offer actionable insights for local governments to design targeted recovery strategies and strengthen community resilience against future public health crises.
0 citationsRead paper
What Matters for Grocery Product Retrieval with Open Source Vision Language Models
This study addresses the challenge of distinguishing highly similar products in fine-grained, zero-shot multimodal SKU retrieval for grocery items. The authors systematically evaluate 190 open-source vision-language models on the GroceryVision benchmark, analyzing the impact of pretraining data quality, model architecture, and input resolution within a contrastive learning embedding and multimodal retrieval framework. Their findings reveal that data quality outweighs scale: lightweight models such as MobileCLIP-B (150M parameters) trained on high-quality data outperform larger models trained on noisier datasets. The work introduces “semantic power density” as a novel efficiency metric. The best-performing model achieves 94.5% Recall@5, yet a 17.5% performance gap remains at Recall@1; notably, data filtering improves accuracy by up to 16.6%.
0 citationsRead paper
Fundamental Limits of 1-bit ISAC Systems: Capacity Region and Optimal Power Control
This study investigates the fundamental trade-off between communication and sensing performance in integrated sensing and communication (ISAC) systems under 1-bit quantization over Gaussian fading channels. Employing information-theoretic tools, the work characterizes the capacity region when channel state information is available at the receiver, revealing that no performance compromise is necessary between communication and sensing under 1-bit quantization. It further demonstrates that rotationally symmetric constant-modulus input distributions simultaneously achieve both communication and sensing capacities. When channel state information is also available at the transmitter, the paper proposes an optimal power control strategy that adaptively adjusts according to sensing priority; the solution continuously transitions from water-filling to uniform power allocation, offering practical design insights for real-world ISAC systems.
0 citationsRead paper
A Survey of Spatial Memory Representations for Efficient Robot Navigation
This study addresses the challenge of unbounded spatial memory growth in visual robotic navigation within large-scale environments, which risks exhausting embedded platform resources. The authors conduct a systematic survey of 52 spatial memory representations from 1989 to 2025 and introduce, for the first time, a memory efficiency metric α—defined as the ratio of runtime memory to map storage—alongside a standardized evaluation protocol. Covering occupancy grids, neural implicit representations, 3D Gaussian Splatting (3DGS), and scene graphs, the analysis integrates GPU performance profiling and memory-completeness curves, revealing that neural methods exhibit α values spanning two orders of magnitude (2.3–215). This indicates that memory architecture, rather than representation paradigm, predominantly governs deployment feasibility. The work releases the first α benchmark dataset and provides α-aware budgeting algorithms with Pareto frontier analysis.
0 citationsRead paper
Recent publications
Latest Papers
FactorFlow: A Visual Analytics Workspace with Large Language Model-Assisted Interpretation for Factor Analysis
为了解决探索性因子分析中模型解释的主观性和复杂性问题，FactorFlow系统通过可视化仪表板和大语言模型辅助生成自然语言解释来帮助研究人员更有效地进行分析。
0 citationsRead paper
Cascading Effects of the COVID-19 Pandemic on Barangays in the Philippines
This study investigates the cascading impacts of the COVID-19 pandemic on Philippine barangays across multiple dimensions, including mobility, public services, economic and financial well-being, food security, education, and physical health. Drawing on survey data from 2,122 households collected between May and June 2021, the research pioneers an integrated approach combining Bayesian networks and influence diagrams, implemented in Python. Variables were subjected to feature clustering and z-score normalization to infer conditional dependencies and potential causal pathways among the multidimensional outcomes. The analysis identifies enhancing food production, stabilizing market prices, and expanding income-generating opportunities as the most effective interventions, underscoring the pivotal role of economic and food security policies. These findings offer actionable insights for local governments to design targeted recovery strategies and strengthen community resilience against future public health crises.
0 citationsRead paper
What Matters for Grocery Product Retrieval with Open Source Vision Language Models
This study addresses the challenge of distinguishing highly similar products in fine-grained, zero-shot multimodal SKU retrieval for grocery items. The authors systematically evaluate 190 open-source vision-language models on the GroceryVision benchmark, analyzing the impact of pretraining data quality, model architecture, and input resolution within a contrastive learning embedding and multimodal retrieval framework. Their findings reveal that data quality outweighs scale: lightweight models such as MobileCLIP-B (150M parameters) trained on high-quality data outperform larger models trained on noisier datasets. The work introduces “semantic power density” as a novel efficiency metric. The best-performing model achieves 94.5% Recall@5, yet a 17.5% performance gap remains at Recall@1; notably, data filtering improves accuracy by up to 16.6%.
0 citationsRead paper
Fundamental Limits of 1-bit ISAC Systems: Capacity Region and Optimal Power Control
This study investigates the fundamental trade-off between communication and sensing performance in integrated sensing and communication (ISAC) systems under 1-bit quantization over Gaussian fading channels. Employing information-theoretic tools, the work characterizes the capacity region when channel state information is available at the receiver, revealing that no performance compromise is necessary between communication and sensing under 1-bit quantization. It further demonstrates that rotationally symmetric constant-modulus input distributions simultaneously achieve both communication and sensing capacities. When channel state information is also available at the transmitter, the paper proposes an optimal power control strategy that adaptively adjusts according to sensing priority; the solution continuously transitions from water-filling to uniform power allocation, offering practical design insights for real-world ISAC systems.
0 citationsRead paper
A Survey of Spatial Memory Representations for Efficient Robot Navigation
This study addresses the challenge of unbounded spatial memory growth in visual robotic navigation within large-scale environments, which risks exhausting embedded platform resources. The authors conduct a systematic survey of 52 spatial memory representations from 1989 to 2025 and introduce, for the first time, a memory efficiency metric α—defined as the ratio of runtime memory to map storage—alongside a standardized evaluation protocol. Covering occupancy grids, neural implicit representations, 3D Gaussian Splatting (3DGS), and scene graphs, the analysis integrates GPU performance profiling and memory-completeness curves, revealing that neural methods exhibit α values spanning two orders of magnitude (2.3–215). This indicates that memory architecture, rather than representation paradigm, predominantly governs deployment feasibility. The work releases the first α benchmark dataset and provides α-aware budgeting algorithms with Pareto frontier analysis.
0 citationsRead paper