arXiv

IDEIA: A Generative AI-Based System for Real-Time Editorial Ideation in Digital Journalism

Leave a Comment / arXiv, Normal Science /

URL https://arxiv.org/pdf/2506.07278.pdf Stage Normal Science Paradigm framing The preprint operates within the contemporary paradigm of digital journalism, which acknowledges the transformative impact of technology on news production and consumption. This paradigm emphasizes the need for speed, thematic relevance, visual appeal, and adaptability in journalistic practices. Within this broader framework, the study specifically addresses the emerging […]

IDEIA: A Generative AI-Based System for Real-Time Editorial Ideation in Digital Journalism Read More »

Happiness Finder: Exploring the Role of AI in Enhancing Well-Being During Four-Leaf Clover Searches

Leave a Comment / arXiv, Model Drift, Normal Science /

URL https://arxiv.org/pdf/2506.07393.pdf Stage Normal Science / Model Drift Paradigm framing The paradigm framing of the preprint is Human-Computer Interaction (HCI) and Positive Computing, specifically focusing on the intersection of AI, well-being, and the experience of search. It operates within the existing paradigm of using technology to enhance human experiences, but with a nuanced approach towards

Happiness Finder: Exploring the Role of AI in Enhancing Well-Being During Four-Leaf Clover Searches Read More »

What is the Quark-Gluon Plasma made of?

Leave a Comment / arXiv, Model Drift /

URL https://arxiv.org/pdf/2506.07181.pdf Stage Model Drift Paradigm framing The current paradigm considers the quark-gluon plasma (QGP) as a strongly coupled liquid-like state composed of short-lived quasiparticles. This preprint evaluates the existing paradigm using both perturbative and non-perturbative methods including hard-thermal loop perturbation theory, lattice QCD, and the functional renormalization group. Highlights This preprint surveys the current

What is the Quark-Gluon Plasma made of? Read More »

NatADiff: Adversarial Boundary Guidance for Natural Adversarial Diffusion

Leave a Comment / arXiv, Normal Science /

URL https://arxiv.org/pdf/2505.20934.pdf Stage Normal Science Paradigm framing The preprint operates within the deep learning paradigm, specifically focusing on adversarial attacks and defenses within the image classification domain. It addresses the sub-paradigm of denoising diffusion probabilistic models for image generation and their application to adversarial sample creation. Highlights The preprint presents a novel method (NatADiff) for

NatADiff: Adversarial Boundary Guidance for Natural Adversarial Diffusion Read More »

Agent Laboratory: Using LLM Agents as Research Assistants

Leave a Comment / arXiv, Model Revolution, Paradigm Change /

URL https://arxiv.org/pdf/2501.04227.pdf Stage Model Revolution / Paradigm Change Paradigm framing The preprint operates within the paradigm of scientific research and publishing, specifically focusing on machine learning subfields. It introduces a potential shift by proposing LLM agents as active participants in the research process. Highlights The preprint "Agent Laboratory: Using LLM Agents as Research Assistants" presents

Agent Laboratory: Using LLM Agents as Research Assistants Read More »

Can LLMs Generate Novel Research Ideas? A Large-Scale Human Study with 100+ NLP Researchers

Leave a Comment / arXiv, Model Drift /

URL https://arxiv.org/pdf/2409.04109.pdf Stage Model Drift Paradigm framing The preprint operates within the established paradigm of using large language models (LLMs) for various Natural Language Processing (NLP) tasks, including idea generation. It specifically addresses the emergent area of research ideation using LLMs, an area still solidifying its own potential paradigms. Highlights This preprint presents a large-scale

Can LLMs Generate Novel Research Ideas? A Large-Scale Human Study with 100+ NLP Researchers Read More »

Shared Imagination: LLMs Hallucinate Alike

Leave a Comment / arXiv, Model Drift /

URL https://arxiv.org/pdf/2407.16604.pdf Stage Model Drift Paradigm framing The preprint operates within the dominant paradigm of natural language processing, specifically focusing on large language models (LLMs). The core assumptions of this paradigm include the effectiveness of transformer-based architectures, the importance of large-scale pre-training data, and the viability of instruction tuning and alignment techniques. Highlights The preprint

Shared Imagination: LLMs Hallucinate Alike Read More »

Exoplaneteers Keep Overestimating Sigma Significances

Leave a Comment / arXiv, Model Drift /

URL https://arxiv.org/pdf/2506.05392.pdf Stage Model Drift Paradigm framing Frequentist and Bayesian statistics in exoplanetary atmospheric studies. Highlights The preprint identifies a drift within the current paradigm of exoplanetary atmospheric characterization. Specifically, it addresses the misapplication of a statistical conversion (Sellke et al. 2001) by exoplaneteers. The authors argue that while the conversion itself is valid, its

Exoplaneteers Keep Overestimating Sigma Significances Read More »

Linear instability of plane Couette and Poiseuille flows

Leave a Comment / arXiv, Model Revolution /

URL https://arxiv.org/pdf/2506.04242.pdf Stage Model Revolution Paradigm framing Hydrodynamic Stability Highlights This preprint challenges the established paradigm in hydrodynamic stability regarding the linear instability of plane Couette and Poiseuille flows. The traditional assumption of longitudinal periodicity of disturbances is questioned, leading to new theoretical predictions for critical Reynolds numbers. The authors provide an alternative model that

Linear instability of plane Couette and Poiseuille flows Read More »

Latent Variable Autoregression with Exogenous Inputs

Leave a Comment / arXiv, Normal Science /

URL https://arxiv.org/pdf/2506.04488.pdf Stage Normal Science Paradigm framing The preprint operates within the dominant paradigm of econometrics and time series analysis, specifically focusing on autoregressive models with exogenous inputs (ARX). It seeks to refine and extend this established model by incorporating latent variables. Highlights This preprint introduces (C)LARX, a novel variant of the ARX methodology. This

Latent Variable Autoregression with Exogenous Inputs Read More »