Thesis Tide - Find relevant and interesting papers

Closed-Form Solutions for Minimum-Time Paths of Dubins Airplane in Steady Wind

This paper is concerned with the minimum-time path-planning problem for a Dubins airplane under the influence of steady wind. The path-planning problem, by transforming into the air-relative frame, is...

Useful Fields:

The article presents a significant advancement in the field of path-planning for Dubins airplanes by providing closed-form solutions that enhance computational efficiency in real-time applications. The application of Pontryagin's maximum principle and the novel approach to categorizing minimum-time solutions enhances methodological rigor. The use of an improved bisection method to find solutions to transcendental equations further strengthens its contributions. Overall, its findings hold potential for broad applicability in air traffic management and autonomous vehicle navigation.

Employee Well-being in the Age of AI: Perceptions, Concerns, Behaviors, and Outcomes

The growing integration of Artificial Intelligence (AI) into Human Resources (HR) processes has transformed the way organizations manage recruitment, performance evaluation, and employee engagement. W...

Useful Fields:

The study addresses the timely and critical intersection of AI and employee well-being, a topic of growing concern as AI becomes more prevalent in workplaces. The novelty lies in the introduction of an Interaction Framework that offers a systematic approach to understanding the dual impact of AI on employee perceptions and outcomes. Methodologically, the paper appears to be based on empirical findings that provide actionable insights for organizations, thus enhancing its applicability. Overall, its implications for human resource practices and employee engagement strategies are highly relevant.

Ternary near-extremal self-dual codes of lengths $36$ , $48$ and $60$

For lengths $36$ , $48$ and $60$ , we construct new ternary near-extremal self-dual codes with weight enumerators for which no ternary near-extremal self-dual codes were previous...

Useful Fields:

The article presents novel constructions of ternary near-extremal self-dual codes, which is a significant addition to the literature as it introduces new examples where none previously existed. The work's methodological rigor is underscored by the focus on specific lengths and the weight enumerators, providing foundational results that could influence future research in coding theory. The novelty and relevance to a prominent topic in the field of coding theory warrant a high relevance score.

Multiplicity of solutions to a degenerate elliptic equation in the sub-critical and critical cases

Given a smooth and bounded domain $Ω(\subset\mathbf{R}^N)$ , we prove the existence of two non-trivial, non-negative solutions for the semilinear degenerate elliptic equation \begin{align} \lef...

Useful Fields:

The article presents novel results on the multiplicity of solutions for a specific class of degenerate elliptic equations, extending the understanding of solution behavior in both sub-critical and critical cases. The use of advanced mathematical frameworks, such as Grushin Laplacian and Sobolev exponents, indicates strong methodological rigor. The implications of this work could significantly impact mathematical physics and variational analysis by providing insights that could lead to further investigations into the non-linear behavior of differential equations.

Distributed Massive MIMO-Aided Task Offloading in Satellite-Terrestrial Integrated Multi-Tier VEC Networks

This paper proposes a distributed massive multiple input multiple-output (DM-MIMO) aided multi-tier vehicular edge computing (VEC) system. In particular, each vehicle terminal (VT) offloads its comput...

Useful Fields:

The article addresses a novel and complex problem of task offloading in vehicular edge computing using a sophisticated DM-MIMO framework, which demonstrates significant potential in optimizing computational efficiency. The methodological rigor is evident in the transformation of non-convex problems into solvable convex problems, suggesting robustness in applied mathematics and operations research. Additionally, the simulation results validate the proposed algorithm's effectiveness, indicating a meaningful contribution to the field.

Multi-class heart disease Detection, Classification, and Prediction using Machine Learning Models

Heart disease is a leading cause of premature death worldwide, particularly among middle-aged and older adults, with men experiencing a higher prevalence. According to the World Health Organization (W...

Useful Fields:

The article presents a novel approach to heart disease detection specifically tailored for the Bangladeshi population, addressing a significant gap in existing research. The introduction of new datasets and the application of advanced machine learning techniques represent both methodological rigor and innovative contributions to the field. The high accuracy achieved has strong implications for clinical practice. The relevance to public health and the potential impact on mortality rates further elevates its significance.

Fault-Tolerant Implementation of the Deutsch-Jozsa Algorithm

The Deutsch-Josza algorithm, one of the first and simplest quantum algorithms, is a natural candidate for small fault-tolerance experiments. We show that one can implement the Deutsch-Josza algorithm ...

Useful Fields:

This article presents a novel implementation of the Deutsch-Jozsa algorithm with a focus on fault tolerance. The use of the $[[4,2,2]]$ error-correcting code represents a significant advancement in quantum computing, particularly for small-scale applications where error resilience is critical. The rigorous methodology and experimental validation on a trapped-ion quantum computer further enhance its credibility and reproducibility. The impressive reduction in error rates (approximately 90%) indicates a practical pathway for improving quantum algorithm performance, which is crucial for the progression of quantum computing technology.

DrIFT: Autonomous Drone Dataset with Integrated Real and Synthetic Data, Flexible Views, and Transformed Domains

Dependable visual drone detection is crucial for the secure integration of drones into the airspace. However, drone detection accuracy is significantly affected by domain shifts due to environmental c...

Useful Fields:

The DrIFT dataset is highly relevant due to its comprehensive approach to addressing the significant challenge of domain shifts in drone detection. The inclusion of both real and synthetic data across multiple domains and conditions enhances its applicability in real-world scenarios. The novel MCDO-map metric represents a methodological advancement that can facilitate further research in the domain adaptation field. Overall, it is both a unique contribution and a practical resource for researchers.

GUIDE: A Global Unified Inference Engine for Deploying Large Language Models in Heterogeneous Environments

Efficiently deploying large language models (LLMs) in real-world scenarios remains a critical challenge, primarily due to hardware heterogeneity, inference framework limitations, and workload complexi...

Useful Fields:

This article addresses a prominent issue in the deployment of large language models by providing a systematic framework that combines dynamic modeling and simulation-based optimization. It is novel due to its focus on tackling performance bottlenecks and its applicability to non-experts, thereby increasing accessibility. Additionally, the methodological rigor in identifying performance issues and detailing the optimization approach strengthens its contribution to the field.

Direct Quantized Training of Language Models with Stochastic Rounding

Although recent quantized Large Language Models (LLMs), such as BitNet, have paved the way for significant reduction in memory usage during deployment with binary or ternary weights, training these mo...

Useful Fields:

The article presents a novel approach to optimizing the training process of quantized language models by implementing stochastic rounding, which can significantly reduce memory requirements while maintaining performance. This innovation addresses a critical challenge in deploying Large Language Models, making it highly relevant for practical applications in AI and machine learning. The empirical results provided enhance the methodological rigor, showcasing effectiveness and potential for future development in low-resource scenarios.

Slicing Vision Transformer for Flexible Inference

Vision Transformers (ViT) is known for its scalability. In this work, we target to scale down a ViT to fit in an environment with dynamic-changing resource constraints. We observe that smaller ViTs ar...

Useful Fields:

The article presents a novel approach to enhance the flexibility of Vision Transformers (ViTs) by proposing a framework that allows a single model to efficiently create multiple sub-networks tailored for varying resource constraints. This adaptability is particularly relevant for deploying ViTs in real-world scenarios where computational resources may fluctuate. Its strong methodological validation and emphasis on practical applications increase its impact potential within the field.

Differentially Private Random Feature Model

Designing privacy-preserving machine learning algorithms has received great attention in recent years, especially in the setting when the data contains sensitive information. Differential privacy (DP)...

Useful Fields:

This article introduces a novel approach to private machine learning by applying differential privacy to random feature models, an area that has not been thoroughly explored before. The methodological rigor in establishing theoretical guarantees and the balance between privacy and performance demonstrates its significance. The concern for disparate impact also aligns with current ethical considerations in AI, further broadening its applicability and relevance.

NLP-ADBench: NLP Anomaly Detection Benchmark

Anomaly detection (AD) is a critical machine learning task with diverse applications in web systems, including fraud detection, content moderation, and user behavior analysis. Despite its significance...

Useful Fields:

The introduction of NLP-ADBench provides a crucial resource for a relatively underexplored area within NLP, specifically anomaly detection. Its comprehensive datasets and exhaustive evaluation of algorithms fill a significant gap and can catalyze future research. The methodological rigor is evident in the detailed comparisons of state-of-the-art models, and the release of the benchmark enhances its applicability and usability within the community. Overall, it represents a critical advancement in both anomaly detection and NLP.

KNN-MMD: Cross Domain Wi-Fi Sensing Based on Local Distribution Alignment

As a key technology in Integrated Sensing and Communications (ISAC), Wi-Fi sensing has gained widespread application in various settings such as homes, offices, and public spaces. By analyzing the pat...

Useful Fields:

The article presents a novel approach (KNN-MMD) to address a significant challenge in the application of Wi-Fi sensing across different environments, a key area within ISAC. The methodology offers enhanced performance through local distribution alignment, demonstrating superiority over traditional DA methods, which is both innovative and practical. The robust evaluation across multiple tasks adds to the credibility and applicability of the findings, suggesting strong potential for future research endeavors.

A Survey of Sustainability in Large Language Models: Applications, Economics, and Challenges

Large Language Models (LLMs) have transformed numerous domains by providing advanced capabilities in natural language understanding, generation, and reasoning. Despite their groundbreaking application...

Useful Fields:

This article addresses the pressing and timely issue of sustainability in the context of rapidly adopted large language models, which is a topic of increasing importance as AI technologies advance. Its comprehensive survey of operational challenges and potential solutions, coupled with a focus on practical applicability for multiple stakeholders, enhances its impact on the field.

DPGIIL: Dirichlet Process-Deep Generative Model-Integrated Incremental Learning for Clustering in Transmissibility-based Online Structural Anomaly Detection

Clustering based on vibration responses, such as transmissibility functions (TFs), is promising in structural anomaly detection, but most existing approaches struggle with determining the optimal clus...

Useful Fields:

The article presents a novel integration of Dirichlet processes with deep generative models for incremental learning in clustering, addressing significant challenges in structural anomaly detection. Its methodological rigor, evidenced by analytical derivation and performance validation against state-of-the-art methods, positions it as highly impactful. The practical applicability in online monitoring scenarios further enhances its relevance.

Anomaly Detection and Classification in Knowledge Graphs

Anomalies such as redundant, inconsistent, contradictory, and deficient values in a Knowledge Graph (KG) are unavoidable, as these graphs are often curated manually, or extracted using machine learnin...

Useful Fields:

The article addresses a pressing issue in knowledge graph maintenance—anomaly detection—using a novel methodology (SEKA) and an associated taxonomy (TAXO). The proposed methods show promising potential for enhancing data quality in KGs, which are crucial for various applications across fields. The evaluation against real-world KGs adds to the methodological rigor, presenting empirical support for the proposed approaches.

Nonlocality-Assisted Enhancement of Error-Free Communication in Noisy Classical Channels

The zero-error capacity of a noisy classical channel quantifies its ability to transmit information with absolute certainty, i.e., without any error. Unlike Shannon's standard channel capacity, wh...

Useful Fields:

The article introduces a novel perspective on error-free communication by leveraging nonlocal correlations, which is a significant advancement beyond classical approaches. The rigorous exploration of the zero-error capacity of noisy classical channels, particularly through well-defined Bell scenarios, indicates strong methodological rigor and originality. Additionally, the implications for communication theory and potential for cross-pollination with quantum information science make this work particularly impactful.

IterNorm: Fast Iterative Normalization

Transformer-based large language models are a memory-bound model whose operation is based on a large amount of data that are marginally reused. Thus, the data movement between a host and accelerator l...

Useful Fields:

The article presents a novel approach to layer normalization specifically tailored for transformer-based large language models, which are pivotal in current AI and machine learning research. The iterative normalization technique proposed is innovative and potentially improves computational efficiency by reducing memory and data transfer requirements. The use of CMOS implementation further emphasizes practical applicability in real-world scenarios, which strengthens its relevance.

Uncovering Vision Modality Threats in Image-to-Image Tasks

Current image generation models can effortlessly produce high-quality, highly realistic images, but this also increases the risk of misuse. In various Text-to-Image or Image-to-Image tasks, attackers ...

Useful Fields:

This article tackles a critical and contemporary issue of security risks associated with image generation models, particularly focusing on threats in the visual modality. The identification of vulnerabilities and the introduction of the VMT-IGMs dataset offers substantial novelty and practical utility, suggesting a clear path for future research to explore and improve defenses against such security threats. The methodology appears rigorous and the relevance of these findings to ongoing discussions about AI ethics and security enhances its impact.