This is a experimental project. Feel free to send feedback!

Thesis Tide

Thesis Tide ranks papers based on their relevance to the fields, with the goal of making it easier to find the most relevant papers. It uses AI to analyze the content of papers and rank them!

DiagramQG: A Dataset for Generating Concept-Focused Questions from Diagrams

Visual Question Generation (VQG) has gained significant attention due to its potential in educational applications. However, VQG researches mainly focus on natural images, neglecting diagrams in educa...

Useful Fields:

The introduction of the DiagramQG dataset fills a significant gap in the field of Visual Question Generation (VQG) by focusing on diagrams in educational contexts, which have been overlooked in prior research. The methodological rigor is highlighted by the introduction of the HKI-DQG framework, which integrates multiple sources of knowledge for improved question generation. The potential educational applications and the utility of the dataset for future research on VQG are substantial, making this work noteworthy.

MTS-UNMixers: Multivariate Time Series Forecasting via Channel-Time Dual Unmixing

Multivariate time series data provide a robust framework for future predictions by leveraging information across multiple dimensions, ensuring broad applicability in practical scenarios. However, thei...

Useful Fields:

The paper presents a novel method (MTS-UNMixer) that addresses significant challenges in multivariate time series forecasting, specifically related to high dimensionality and interpretability. The dual unmixing approach is innovative, and the use of both historical and future data to enhance predictions adds to its rigor. The reported experimental results demonstrate substantial improvements over existing techniques, suggesting both practical applicability and potential for further research development.

Omegance: A Single Parameter for Various Granularities in Diffusion-Based Synthesis

In this work, we introduce a single parameter $ω$ , to effectively control granularity in diffusion-based synthesis. This parameter is incorporated during the denoising steps of the diffusion m...

Useful Fields:

The introduction of the single parameter $ω$ for controlling granularity in diffusion-based synthesis represents a novel contribution that enhances the flexibility and applicability of existing models without the need for retraining. Its potential for improving the quality of generated outputs across multiple image and video synthesis tasks demonstrates clear methodological rigor and practical significance. The ability to facilitate region-specific or timestep-specific control adds an exciting dimension to the work, likely inspiring future research in related fields.

Graph Neural Network for Cerebral Blood Flow Prediction With Clinical Datasets

Accurate prediction of cerebral blood flow is essential for the diagnosis and treatment of cerebrovascular diseases. Traditional computational methods, however, often incur significant computational c...

Useful Fields:

This article presents a novel application of graph neural networks (GNNs) in predicting cerebral blood flow, which is critical for cerebrovascular disease diagnosis and treatment. The use of clinical datasets and the model's ability to generalize to unseen vascular structures are significant advancements. The strong performance metrics (Pearson’s correlations) indicate methodological rigor and practical relevance, providing a substantial contribution to both computational neuroscience and clinical applications.

Barriers to Adopting Design for Assembly in Modular Product Architecture: Development of a Conceptual Model Through Content Analysis

This study investigates the barriers to integrating Design for Assembly (DFA) principles within modular product architectures established using the Modular Function Deployment (MFD) method -- a critic...

Useful Fields:

The article presents a novel investigation into barriers of Design for Assembly (DFA) in modular product architecture, reflecting a pressing issue as industries shift toward mass customization. Its methodological rigor through qualitative content analysis involving expert perspectives enhances its reliability. The newly developed conceptual model offers practical solutions for overcoming barriers, potentially influencing industry practices. The interdisciplinary approach emphasizes collaboration, which is essential for practical implementation, highlighting broader implications across related fields.

Exploring Aleatoric Uncertainty in Object Detection via Vision Foundation Models

Datasets collected from the open world unavoidably suffer from various forms of randomness or noiseness, leading to the ubiquity of aleatoric (data) uncertainty. Quantifying such uncertainty is partic...

Useful Fields:

This article addresses a crucial issue in object detection by introducing a novel approach to quantify aleatoric uncertainty using vision foundation models. The proposed methodology not only reveals a deeper understanding of underlying uncertainties but also suggests practical applications that enhance training robustness, making it potentially transformative for the field. The methodological rigor is evident in the empirical validation across multiple advanced models and benchmarks, underscoring its applicability.

Integrating Dual Prototypes for Task-Wise Adaption in Pre-Trained Model-Based Class-Incremental Learning

Class-incremental learning (CIL) aims to acquire new classes while conserving historical knowledge incrementally. Despite existing pre-trained model (PTM) based methods performing excellently in CIL, ...

Useful Fields:

This article presents a novel approach to class-incremental learning (CIL) through the introduction of the Dual Prototype network, which addresses two significant issues: catastrophic forgetting during fine-tuning of pre-trained models and enhancing task adaptation capabilities. The methodological rigor is notable, and the empirical results suggest substantial improvement over existing methods, indicating strong applicability to real-world scenarios. The open-sourcing of code adds further value for the community.