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!

Deriving Activation Functions via Integration

Activation functions play a crucial role in introducing non-linearities to deep neural networks. We propose a novel approach to designing activation functions by focusing on their gradients and derivi...

Useful Fields:

The article introduces a novel approach to the design of activation functions in neural networks, which is a crucial aspect of improving model performance. Its methodological rigor is highlighted by the empirical results showing the effectiveness of the proposed xIELU function compared to established functions. The focus on gradients and integration for activation functions is innovative and could inspire new lines of research in neural network architectures.

Evaluating LLMs Capabilities Towards Understanding Social Dynamics

Social media discourse involves people from different backgrounds, beliefs, and motives. Thus, often such discourse can devolve into toxic interactions. Generative Models, such as Llama and ChatGPT, h...

Useful Fields:

The article discusses the crucial ability of large language models (LLMs) to interpret social dynamics, which is a growing area of interest. The focus on social media interactions, particularly in the context of cybersecurity and cyberbullying, makes it highly relevant, with potential applications for technology design and social research. The comparative analysis of different LLMs also contributes to a deeper understanding of their capabilities and limitations, an aspect that is essential for future research.

Automating Sonologists USG Commands with AI and Voice Interface

This research presents an advanced AI-powered ultrasound imaging system that incorporates real-time image processing, organ tracking, and voice commands to enhance the efficiency and accuracy of diagn...

Useful Fields:

This article presents a highly innovative approach to ultrasound imaging by integrating AI and voice command functionalities, which addresses the issues of time consumption and subjectivity inherent in traditional diagnostics. The application of advanced techniques like Mask R-CNN for semantic segmentation and the achievement of high accuracy in liver histopathology showcases methodological rigor and significant potential for improving clinical outcomes. Furthermore, the real-time processing aspect adds considerable value for practical applications. Its interdisciplinary nature, combining AI, medical imaging, and voice recognition, reinforces its impact and relevance in various fields.

DT-LSD: Deformable Transformer-based Line Segment Detection

Line segment detection is a fundamental low-level task in computer vision, and improvements in this task can impact more advanced methods that depend on it. Most new methods developed for line segment...

Useful Fields:

The article presents a novel approach to line segment detection using deformable transformer-based models, which is a significant advancement over traditional CNN-based methods. Its introduction of Line Contrastive DeNoising (LCDN) not only enhances model training efficiency but also pushes the accuracy boundaries in this essential computer vision task. This combination of innovation in model architecture and methodological improvement suggests a strong potential for impact in the field.

MERLOT: A Distilled LLM-based Mixture-of-Experts Framework for Scalable Encrypted Traffic Classification

We present MERLOT, a scalable mixture-of-expert (MoE) based refinement of distilled large language model optimized for encrypted traffic classification. By applying model distillation techniques in a ...

Useful Fields:

This article introduces a novel framework (MERLOT) that combines large language model distillation with Mixture-of-Experts architecture, addressing a critical challenge in encrypted traffic classification with a focus on scalability and efficiency. The methodological rigor is evident through comprehensive experimentation across multiple datasets, demonstrating superior performance compared to existing models. The applicability is broad as it tackles an urgent problem in cybersecurity, making it highly relevant for future research and practical applications.

The Alexander polynomial of twisted torus knots

Twisted torus knots are a generalization of torus knots, obtained by introducing additional full twists to adjacent strands of the torus knots. In this article, we present an explicit formula for the ...

Useful Fields:

The article presents a novel formula for the Alexander polynomial of twisted torus knots, which is an important contribution to knot theory and could drive further investigation into related knot invariants. Utilizing Fox's free differential calculus adds methodological rigor, suggesting that the results are robust. Furthermore, the implications for knot genus and $L$-space knots highlight potential applications in understanding the topology of three-manifolds, which positions this work as influential for future studies in this area.

Multi-scale simulation of red blood cell trauma in large-scale high-shear flows after Norwood operation

Cardiovascular surgeries and mechanical circulatory support devices create non-physiological blood flow conditions that can be detrimental, especially for pediatric patients. A source of complications...

Useful Fields:

The study presents a novel multi-scale numerical model that addresses a critical problem in pediatric cardiovascular surgeries. Its methodological rigor in using CFD and Lagrangian tracking provides robust insights into the hemolytic effects of surgical shunt designs, which is highly relevant for improving patient outcomes. The results offer substantial clinical implications and could inspire future research on blood flow dynamics and surgical design optimization.

Collaborative Feature-Logits Contrastive Learning for Open-Set Semi-Supervised Object Detection

Current Semi-Supervised Object Detection (SSOD) methods enhance detector performance by leveraging large amounts of unlabeled data, assuming that both labeled and unlabeled data share the same label s...

Useful Fields:

The article presents a novel approach to improving semi-supervised object detection in open-set scenarios, which is a significant challenge in current research. The introduction of a feature-level clustering method with contrastive loss and the optimization of logits-level uncertainties represent robust methodological advancements. The extensive experiments validating the proposed method further enhance its credibility and potential impact in the field.

NCAirFL: CSI-Free Over-the-Air Federated Learning Based on Non-Coherent Detection

Over-the-air federated learning (FL), i.e., AirFL, leverages computing primitively over multiple access channels. A long-standing challenge in AirFL is to achieve coherent signal alignment without rel...

Useful Fields:

The proposed NCAirFL scheme addresses a significant challenge in over-the-air federated learning by eliminating the need for channel state information, thus enhancing efficiency and practicality in real-world applications. The novelty lies in its approach of using non-coherent detection and error compensation, and its quantitative results indicating improved convergence rates suggest strong methodological rigor. The relevance of this work lies in addressing current technological limitations, making it potentially transformative for federated learning in resource-constrained environments.

From Signal Space To STP-CS

Under the assumption that a finite signal with different sampling lengths or different sampling frequencies is considered as equivalent, the signal space is considered as the quotient space of $\m...

Useful Fields:

The article presents a novel approach to conceptualizing signal space through the lens of quotient spaces, which can lead to innovative insights in compressed sensing techniques. Its rigorous exploration of topological structures and the application to sensing matrix construction adds to its relevance. The combination of theoretical foundations and practical applications enhances its potential impact on the field.

Gracefulness of two nested cycles: a first approach

It is known that if a plane graph is graceful (resp. near-graceful), then its semidual is conservative (resp. near-conservative). In this work we prove that the semidual of a plane graph of size $...

Useful Fields:

The article addresses specific properties of planar graphs related to gracefulness, which is a relatively niche topic. While it presents useful mathematical proofs, the applicability of its results outside of theoretical graph theory may be limited. However, it does contribute towards understanding the attributes of semidual graphs, which could foster further inquiry in this area. The methodological rigor appears sound, yet the novelty of the findings may not be substantial enough to draw broad interest across various related fields.

Gravitational entanglement witness through Einstein ring image

We investigate the interplay between quantum theory and gravity by exploring gravitational lensing and Einstein ring images in a weak gravitational field induced by a mass source in spatial quantum su...

Useful Fields:

The article poses a novel exploration of the intersection between quantum mechanics and general relativity, specifically through the lens of gravitational lensing and Einstein ring images. Its emphasis on visualizing gravitational entanglement offers a fresh avenue of research that can significantly impact both quantum gravity studies and experimental methodologies in cosmology. The methodological rigor followed in analyzing distinct gravitational models adds robustness, while the introduction of specific metrics such as the which-path information indicator helps to bridge theory with observable phenomena.

Asymptotics in Wasserstein Distance for Empirical Measures of Markov Processes

In this paper we introduce some recent progresses on the convergence rate in Wasserstein distance for empirical measures of Markov processes. For diffusion processes on compact manifolds possibly with...

Useful Fields:

This article presents significant advancements in understanding the convergence of empirical measures for Markov processes, particularly in terms of their Wasserstein distance. Its methodological rigor, particularly the establishment of a precise convergence rate and the use of renormalization limits, is noteworthy. Such findings can influence a range of applications in stochastic processes and probability theory, marking the work as both impactful and relevant.

Eliminating Ratio Bias for Gradient-based Simulated Parameter Estimation

This article addresses the challenge of parameter calibration in stochastic models where the likelihood function is not analytically available. We propose a gradient-based simulated parameter estimati...

Useful Fields:

The article introduces a novel gradient-based simulated parameter estimation framework that addresses a significant gap in the calibration of stochastic models. Its theoretical contributions regarding strong convergence and asymptotic normality provide a solid foundation for further research, while its practical applicability to neural networks indicates widespread relevance across fields. The proposed algorithm's ability to optimize accuracy and computational efficiency strengthens its impact.

Revisiting the activity-rotation relation for evolved stars

The magnetic dynamo mechanism of giant stars remains an open question, which can be explored by investigating their activity-rotation relations with multiple proxies. By using the data from the LAMOST...

Useful Fields:

The article presents a comprehensive investigation into the activity-rotation relations of evolved stars, employing a variety of data which enhances methodological rigor. The findings suggest fundamental understanding of stellar activity that bridges gaps between different stellar classifications, indicating significance and novelty in its contributions to current astrophysical theories on magnetic dynamo mechanisms in stars. This could yield substantial implications for stellar evolution models and our understanding of stellar populations.

HapKnob -- A Motorized Shape-changing Haptic Knob Interface

The absence of physical interfaces creates challenges when interacting with touchscreen technology. This study aims to investigate an innovative haptic solution for interacting with graphical user int...

Useful Fields:

The article presents a novel motorized haptic knob interface that addresses a significant challenge in touchscreen technology interaction. The innovative design of HapKnob, with its multiple shape configurations and force feedback capabilities, offers a fresh approach that pushes the boundaries of user interface design. Its potential applicability in environments with limited visual feedback increases its relevance and could inspire further research into multi-sensory interfaces.

MemoryFormer: Minimize Transformer Computation by Removing Fully-Connected Layers

In order to reduce the computational complexity of large language models, great efforts have been made to to improve the efficiency of transformer models such as linear attention and flash-attention. ...

Useful Fields:

MemoryFormer presents a novel architectural design that significantly reduces computational complexity without sacrificing performance. Its innovative approach to replacing fully-connected layers is highly relevant in the context of large language models, where efficiency is paramount. The empirical validation across various benchmarks strengthens its impact, suggesting a solid methodological foundation and potential for widespread adoption.

Resolving phonon-mediated superconducting pairing symmetries from first-principles calculation

Thequest for topological superconductors triggers revived interests in resolving non-s-wave pairing channels mediated by phonons. While density functional theory and denstify functional perturbtaion t...

Useful Fields:

This article presents a novel algorithm that extends the applicability of first-principles calculations to non-s-wave superconductivity, a relatively underexplored area. The methodological innovation and its application to real materials could significantly advance our understanding of topological superconductors and promote further exploration in this field.

BetterBench: Assessing AI Benchmarks, Uncovering Issues, and Establishing Best Practices

AI models are increasingly prevalent in high-stakes environments, necessitating thorough assessment of their capabilities and risks. Benchmarks are popular for measuring these attributes and for compa...

Useful Fields:

This article tackles critical issues regarding the quality and usability of AI benchmarks, which are fundamental for the reliable assessment of AI models in important applications. Its comprehensive framework and practical checklist for benchmark developers represent significant advancements in the field. The implications for standardization and quality assurance are profound, which can lead to improved practices across many areas involved in AI.

Data Watermarking for Sequential Recommender Systems

In the era of large foundation models, data has become a crucial component for building high-performance AI systems. As the demand for high-quality and large-scale data continues to rise, data copyrig...

Useful Fields:

This article addresses a novel approach to data watermarking specifically for sequential recommender systems, which is highly relevant given the increasing concern for data copyright in AI applications. The methodology presented is rigorous, with systematic definitions of the watermarking problems and extensive experimental validation. Furthermore, its practical applicability to high-performance AI systems makes this work stand out.