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!

CGI: Identifying Conditional Generative Models with Example Images

Generative models have achieved remarkable performance recently, and thus model hubs have emerged. Existing model hubs typically assume basic text matching is sufficient to search for models. However,...

Useful Fields:

The article presents a novel approach to identifying conditional generative models using user-provided example images, which addresses a significant gap in existing model hubs that rely primarily on text-based searches. The proposed methodology, PromptBased Model Identification (PMI), showcases methodological rigor with empirical evaluations against a substantial benchmark of models and tasks. This innovation has strong implications for improving user efficiency and accessibility in model selection, representing an important advancement in the field of generative models.

A visual representation of the properties of pre- and post- selected entangled systems

We introduce a visual representation for generating entangled-based quantum effects under pre- and post- selected states that allows us to reveal equivalence between seemingly different quantum effect...

Useful Fields:

The article presents novel visual representations for complex quantum effects related to entangled states, addressing an important aspect of quantum mechanics that can unify several existing concepts. Its methodological rigor and focus on generalizing key paradoxes in quantum theory demonstrate significant advancement in understanding high-dimensional entangled systems. This has the potential to inspire future experiments and research paradigms in quantum theory, which is crucial for the advancement of quantum computing and quantum information science.

FreEformer: Frequency Enhanced Transformer for Multivariate Time Series Forecasting

This paper presents \textbf{FreEformer}, a simple yet effective model that leverages a \textbf{Fre}quency \textbf{E}nhanced Trans\textbf{former} for multivariate time series forecasting. Our work is b...

Useful Fields:

This article presents a novel approach (FreEformer) that significantly enhances Time Series Forecasting using a Frequency Enhanced Transformer. The use of the frequency domain for representation learning is innovative and can lead to improved performance in forecasting tasks. The strong empirical results across diverse domains further strengthen its applicability.

MCRL4OR: Multimodal Contrastive Representation Learning for Off-Road Environmental Perception

Most studies on environmental perception for autonomous vehicles (AVs) focus on urban traffic environments, where the objects/stuff to be perceived are mainly from man-made scenes and scalable dataset...

Useful Fields:

The article presents a novel approach to environmental perception in off-road settings, which is significantly underexplored compared to urban environments. The research employs a multimodal contrastive learning framework that integrates various data types, showcasing methodological rigor and promising results in performance benchmarks. This originality, coupled with its applicability to the emerging field of off-road autonomous vehicles, indicates substantial potential for both meaningful advancements in the field and inspiration for future research.

OstQuant: Refining Large Language Model Quantization with Orthogonal and Scaling Transformations for Better Distribution Fitting

Post-training quantization (PTQ) has emerged as a widely adopted technique for compressing and accelerating Large Language Models (LLMs). The major challenge in LLM quantization is that uneven and hea...

Useful Fields:

The article presents a novel approach to quantizing large language models using quantitative metrics and mathematical derivations, providing a rigorous foundation for its methodology. The inclusion of the Quantization Space Utilization Rate (QSUR) and the KL-Top loss function reflects significant innovation in addressing limitations in existing quantization techniques. The empirical results showing superior performance further underscore its potential impact on the optimization of model efficiency and accuracy, making it highly relevant for advancing research in LLM quantization techniques.

Pilot: Building the Federated Multimodal Instruction Tuning Framework

In this paper, we explore a novel federated multimodal instruction tuning task(FedMIT), which is significant for collaboratively fine-tuning MLLMs on different types of multimodal instruction data on ...

Useful Fields:

The article introduces a novel framework for federated multimodal instruction tuning, addressing significant challenges in collaborative learning with distributed data. Its innovation lies in the effective integration of adaptation mechanisms for both task-specific and client-specific features, demonstrating methodological rigor and potential for broad applicability in real-world scenarios. The proposal of an adaptive parameter aggregation strategy enhances its robustness, making this work highly relevant to the field.

Comprehensive Modeling and Question Answering of Cancer Clinical Practice Guidelines using LLMs

The updated recommendations on diagnostic procedures and treatment pathways for a medical condition are documented as graphical flows in Clinical Practice Guidelines (CPGs). For effective use of the C...

Useful Fields:

This article presents a novel approach that combines advanced LLMs with a graph-based representation of cancer clinical practice guidelines, addressing a significant gap in the current literature on Clinical Decision Support Systems. By effectively enhancing the interpretation of CPGs, the methodology holds potential for improving patient treatment decisions and outcomes. The accuracy achieved in node classification further demonstrates methodological rigor and applicability to real-world healthcare scenarios.

AdEval: Alignment-based Dynamic Evaluation to Mitigate Data Contamination in Large Language Models

As Large Language Models (LLMs) are pretrained on massive-scale corpora, the issue of data contamination has become increasingly severe, leading to potential overestimation of model performance during...

Useful Fields:

The article introduces a novel method, AdEval, which addresses a significant issue in LLM evaluation -- data contamination. The approach is methodologically rigorous, leveraging dynamic evaluation that aligns with cognitive frameworks. Its practical implications for improving evaluation reliability are substantial, particularly in refining performance measures for LLMs, a vital aspect in AI research and development. This innovative framework is likely to spur further research in evaluation methodologies and the integrity of data used in machine learning.

Attribute-based Visual Reprogramming for Image Classification with CLIP

Visual reprogramming (VR) reuses pre-trained vision models for downstream image classification tasks by adding trainable noise patterns to inputs. When applied to vision-language models (e.g., CLIP), ...

Useful Fields:

The article presents a novel approach to visual reprogramming that enhances the performance of vision-language models by employing attribute-based representations. The use of descriptive and distinctive attributes provides a new dimension to optimizing image classification, potentially addressing limitations in existing label-based techniques. Its empirical validation across various downstream tasks affirms its applicability and robustness, making it a valuable contribution to both the fields of machine learning and computer vision.

The ICEBERG Test Stand for DUNE Cold Electronics Development

ICEBERG is a liquid argon time projection chamber at Fermilab for the purpose of testing detector components and software for the Deep Underground Neutrino Experiment (DUNE). The detector features a 1...

Useful Fields:

The ICEBERG Test Stand addresses critical technological advancements needed for the DUNE experiment, particularly in testing and optimizing electronics for neutrino detection. The combination of traditional detector techniques with AI-based analysis demonstrates both novelty and methodological rigor, indicating a forward-thinking approach that can influence future detector designs. However, the reliance on experimental setups that are in progress limits the immediate applicability of some findings.

A Semiparametric Bayesian Method for Instrumental Variable Analysis with Partly Interval-Censored Time-to-Event Outcome

This paper develops a semiparametric Bayesian instrumental variable analysis method for estimating the causal effect of an endogenous variable when dealing with unobserved confounders and measurement ...

Useful Fields:

This article presents a novel semiparametric Bayesian method that addresses the complex issue of instrumental variable analysis in the presence of partly interval-censored time-to-event data. Its methodological rigor, particularly with the development of a two-stage Dirichlet process mixture model, enhances its relevance by offering robust alternatives to traditional analysis techniques, making it significant for causal inference studies. The application of the method to a real dataset (UK Biobank) further demonstrates its practical implications in epidemiology, which underlines its potential impact on future research developments in related fields.

Symbolic Knowledge Extraction and Injection with Sub-symbolic Predictors: A Systematic Literature Review

In this paper we focus on the opacity issue of sub-symbolic machine learning predictors by promoting two complementary activities, namely, symbolic knowledge extraction (SKE) and injection (SKI) from ...

Useful Fields:

This article addresses a critical issue in the field of artificial intelligence—namely, the opacity of sub-symbolic predictors, which is essential for enhancing explainability in machine learning models. The systematic review of methods, alongside proposed taxonomies, offers significant contributions that could guide future research and development in explainable AI (XAI). The robust methodological approach and the inclusion of runnable software implementations strengthen its utility and applicability in practical scenarios.

A plethora of long-range neutrino interactions probed by DUNE and T2HK

The next-generation neutrino oscillation experiments would be sensitive to the new neutrino interactions that would strengthen the search for physics beyond the Standard Model. In this context, we exp...

Useful Fields:

The article addresses novel long-range neutrino interactions and their potential implications for physics beyond the Standard Model, making significant strides in the investigation of neutrino properties and interactions. The combination of DUNE and T2HK experiments with new theoretical frameworks offers a fresh perspective that could advance both experimental and theoretical particle physics. The methodological rigor in forecasting sensitivities adds robustness to the findings.

Quasinormal frequencies in Reissner-Nordström de Sitter black holes: constraints from space-time and scalar field parameters

We examine the quasinormal modes exhibited by a massive scalar test field carrying an electric charge, oscillating in the outer region of a Reissner-Nordström de Sitter black hole. We examine the quas...

Useful Fields:

This article presents a novel investigation into the quasinormal modes of charged scalar fields in the context of Reissner-Nordström de Sitter black holes, which is an important extension of previous work in black hole physics. The use of a semi-classical method adds methodological rigor, and the findings regarding the effective potential and its relation to the scalar field parameters contribute new insights into black hole thermodynamics and stability. The implications for astrophysical phenomena and gravitational wave astronomy are particularly significant, warranting a high relevance score.

Heliometric stereo: a new frontier in surface profilometry

Accurate and reliable measurements of three-dimensional surface structures are important for a broad range of technological and research applications, including materials science, nanotechnology, and ...

Useful Fields:

This article presents a novel technique (heliometric stereo) that enhances the quantitative measurement capabilities of SHeM in surface profilometry, which is a significant advancement in a crucial area of research. The method shows promise for accuracy and provides practical insights for implementation, increasing its relevance. The interdisciplinary nature of the subject contributes to broader applicability in multiple fields.

Resource Allocation Driven by Large Models in Future Semantic-Aware Networks

Large model has emerged as a key enabler for the popularity of future networked intelligent applications. However, the surge of data traffic brought by intelligent applications puts pressure on the re...

Useful Fields:

The article presents a novel approach to resource allocation in semantic-aware networks, leveraging large models for efficient data transmission. The integration of scene graph and multimodal pre-trained models indicates methodological rigor and relevance. Its focus on semantic transmission quality as a key metric addresses contemporary challenges in network efficiency and energy consumption, promising significant advancements in future intelligent applications.

Radial Uncertainty Product for Spherically Symmetric Potential in Position Space

This paper presents a detailed analysis of the radial uncertainty product for quantum systems with spherically symmetric potentials. Using the principles of quantum mechanics, the study derives the ra...

Useful Fields:

The article contributes novel insights into the radial uncertainty product within quantum mechanics, particularly for spherically symmetric potentials. Its methodological rigor, including both analytical and numerical approaches, enhances the understanding of quantum systems, opening pathways for future exploration in quantum mechanics. The focus on specific systems like the Hydrogen atom adds practical applicability, though the confinement to non-relativistic regimes may limit broader applicability to relativistic quantum systems.