Tootfinder

Extended AI Interactions Shape Sycophancy and Perspective Mimesis
Shomik Jain, Charlotte Park, Matheus Mesquita Viana, Ashia Wilson, Dana Calacci
https://arxiv.org/abs/2509.12517

Extended AI Interactions Shape Sycophancy and Perspective Mimesis
We investigate whether long-context interactions between users and LLMs lead to AI mirroring behaviors. We focus on two forms of mirroring: (1) sycophancy -- the tendency of models to be overly agreeable with users, and (2) perspective mimesis -- the extent to which models reflect a user's perspective. Using two weeks of interaction context collected from 38 users, we compare model responses with and without long-context for two tasks: political explanations and personal advice. Our results dem…

Humanoid Robot Acrobatics Utilizing Complete Articulated Rigid Body Dynamics
Gerald Brantner
https://arxiv.org/abs/2508.08258 https://arxiv.org/pdf/2508.08…

Humanoid Robot Acrobatics Utilizing Complete Articulated Rigid Body Dynamics
Endowing humanoid robots with the ability to perform highly dynamic motions akin to human-level acrobatics has been a long-standing challenge. Successfully performing these maneuvers requires close consideration of the underlying physics in both trajectory optimization for planning and control during execution. This is particularly challenging due to humanoids' high degree-of-freedom count and associated exponentially scaling complexities, which makes planning on the explicit equations of motio…

Energy-Aware Code Generation with LLMs: Benchmarking Small vs. Large Language Models for Sustainable AI Programming
Humza Ashraf, Syed Muhammad Danish, Aris Leivadeas, Yazan Otoum, Zeeshan Sattar
https://arxiv.org/abs/2508.08332

Energy-Aware Code Generation with LLMs: Benchmarking Small vs. Large Language Models for Sustainable AI Programming
Large Language Models (LLMs) are widely used for code generation. However, commercial models like ChatGPT require significant computing power, which leads to high energy use and carbon emissions. This has raised concerns about their environmental impact. In this study, we evaluate open-source Small Language Models (SLMs) trained explicitly for code generation and compare their performance and energy efficiency against large LLMs and efficient human-written Python code. The goal is to investigat…

SPARE: Securing Progressive Web Applications Against Unauthorized Replications
Sajib Talukder, Nur Imtiazul Haque, Khandakar Ashrafi Akbar
https://arxiv.org/abs/2508.07053 https…

SPARE: Securing Progressive Web Applications Against Unauthorized Replications
WebView applications are widely used in mobile applications to display web content directly within the app, enhancing user engagement by eliminating the need to open an external browser and providing a seamless experience. Progressive Web Applications (PWAs) further improve usability by combining the accessibility of web apps with the speed, offline capabilities, and responsiveness of native applications. However, malicious developers can exploit this technology by duplicating PWA web links to …

Video-Based MPAA Rating Prediction: An Attention-Driven Hybrid Architecture Using Contrastive Learning
Dipta Neogi, Nourash Azmine Chowdhury, Muhammad Rafsan Kabir, Mohammad Ashrafuzzaman Khan
https://arxiv.org/abs/2509.06826

Video-Based MPAA Rating Prediction: An Attention-Driven Hybrid Architecture Using Contrastive Learning
The rapid growth of visual content consumption across platforms necessitates automated video classification for age-suitability standards like the MPAA rating system (G, PG, PG-13, R). Traditional methods struggle with large labeled data requirements, poor generalization, and inefficient feature learning. To address these challenges, we employ contrastive learning for improved discrimination and adaptability, exploring three frameworks: Instance Discrimination, Contextual Contrastive Learning, …

Inoculation Prompting: Instructing LLMs to misbehave at train-time improves test-time alignment
Nevan Wichers, Aram Ebtekar, Ariana Azarbal, Victor Gillioz, Christine Ye, Emil Ryd, Neil Rathi, Henry Sleight, Alex Mallen, Fabien Roger, Samuel Marks
https://arxiv.org/abs/2510.05024

Inoculation Prompting: Instructing LLMs to misbehave at train-time improves test-time alignment
Large language models are sometimes trained with imperfect oversight signals, leading to undesired behaviors such as reward hacking and sycophancy. Improving oversight quality can be expensive or infeasible, motivating methods that improve learned behavior despite an imperfect training signal. We introduce Inoculation Prompting (IP), a simple but counterintuitive technique that prevents learning of an undesired behavior by modifying training prompts to explicitly request it. For example, to ino…

Stabilization techniques for immersogeometric analysis of plate and shell problems in explicit dynamics
Giuliano Guarino, Yannis Voet, Pablo Antolin, Annalisa Buffa
https://arxiv.org/abs/2509.00522

Stabilization techniques for immersogeometric analysis of plate and shell problems in explicit dynamics
Finite element plate and shell formulations are ubiquitous in structural analysis for modeling all kinds of slender structures, both for static and dynamic analyses. The latter are particularly challenging as the high order nature of the underlying partial differential equations and the slenderness of the structures all impose a stringent constraint on the critical time step in explicit dynamics. Unfortunately, badly cut elements in immersed finite element discretizations further aggravate the …

Exploring cosmological constraints on galaxy formation time
Agripino Sousa-Neto, Maria Aldin\^ez Dantas, Javier E. Gonz\'alez, Joel C. Carvalho, Jailson Alcaniz
https://arxiv.org/abs/2509.26611

Exploring cosmological constraints on galaxy formation time
The Universe consists of a variety of objects that formed at different epochs, leading to variations in the formation time which represents the time elapsed from the onset of structure formation until the formation time of a particular object. In this work, we present two approaches to reconstruct and constrain the galaxy formation time $t_f(z)$ using non-parametric reconstruction methods, such as Gaussian Processes (GP) and High-performance Symbolic Regression (SR). Our analysis uses age estim…

Learning Real-World Acrobatic Flight from Human Preferences
Colin Merk, Ismail Geles, Jiaxu Xing, Angel Romero, Giorgia Ramponi, Davide Scaramuzza
https://arxiv.org/abs/2508.18817

Learning Real-World Acrobatic Flight from Human Preferences
Preference-based reinforcement learning (PbRL) enables agents to learn control policies without requiring manually designed reward functions, making it well-suited for tasks where objectives are difficult to formalize or inherently subjective. Acrobatic flight poses a particularly challenging problem due to its complex dynamics, rapid movements, and the importance of precise execution. In this work, we explore the use of PbRL for agile drone control, focusing on the execution of dynamic maneuve…

GATEBLEED: Exploiting On-Core Accelerator Power Gating for High Performance & Stealthy Attacks on AI
Joshua Kalyanapu, Farshad Dizani, Darsh Asher, Azam Ghanbari, Rosario Cammarota, Aydin Aysu, Samira Mirbagher Ajorpaz
https://arxiv.org/abs/2507.17033

GATEBLEED: Exploiting On-Core Accelerator Power Gating for High Performance & Stealthy Attacks on AI
As power consumption from AI training and inference continues to increase, AI accelerators are being integrated directly into the CPU. Intel's Advanced Matrix Extensions (AMX) is one such example, debuting on the 4th generation Intel Xeon Scalable CPU. We discover a timing side and covert channel, GATEBLEED, caused by the aggressive power gating utilized to keep the CPU within operating limits. We show that the GATEBLEED side channel is a threat to AI privacy as many ML models such as transform…