Traditional machine learning relies on explicit models and domain assumptions, often limiting flexibility and interpretability. We introduce a "model-free" framework of instance-based learning using surprisal (information theoretic uncertainty) to directly analyze and perform inferences from raw data, eliminating distr...
Christopher J. Hazard, Michael Resnick, Jacob Beel et al.· 0 citations
This paper shows that standard (post-hoc) calibrators can be trained end-to-end with the backbone via a bilevel Stackelberg formulation, and shows that using the final classification layer itself as the calibrator yields accurate, well-calibrated models with no additional parameters.
Nektarios G. Kalampalikis, Pablo M. Olmos, Isabel Valera· 0 citations
Linearizing softmax attention offers a route to efficient Transformer inference, but preserving pretrained attention remains challenging. We identify an important source of approximation error overlooked by standard linearization methods: a small number of outlier dimensions contribute disproportionately to query-key d...
Wei-Qi Sun, Qian-Yue Wu, Wen-Jia Wang et al.· 0 citations
When a language model is distilled from a teacher that shares its base parameters, the student can inherit behavioural traits of the teacher---including misalignment---even from data that is semantically unrelated to the trait (Cloud et al., 2026). All existing defences require knowing the trait. We remove this require...
Joseph Egan, C. Gilligan-Lee, Michael O'Riordan et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
Explainable Artificial Intelligence (XAI) tools have often been presented as methods to increase trust in machine learning models by making aspects of their algorithms more human-understandable. However, evidence accumulated over the last decade points to an overall failure of this objective, instead promoting overreli...
Marco Zullich, Emily Schiller, Ivan Gentile et al.· 0 citations
Official Scientific Intelligence Gateway: https://sinotechintel.com This repository provides a comprehensive, structured open dataset indexing 1,381 authentic Chinese scientific research publications in Advanced Metallurgy, High-Entropy Alloys, Metal 3D Printing, Titanium & Deep Tech. Curated and verified by the SinoTe...
SinoTechIntel Academic Editorial Office· Zenodo (CERN European Organi...· 0 citations
Background and Rationale Malignant ovarian germ cell tumours (MOGCTs) are uncommon ovarian malignancies that predominantly affect children, adolescents, and young adults. Fertility-sparing surgery combined with platinum-based chemotherapy is an important treatment approach for eligible patients, aiming to achieve oncol...
Jeevarathinam Thirumalai· Open Science Framework· 0 citations
Science, engineering, and health-related laboratories in tertiary institutions generate liquid waste that may contain toxic heavy metals, including lead, cadmium, chromium, manganese, and iron. Poor segregation, treatment, handling, and disposal of laboratory waste may contaminate drains and nearby water sources. This...
Sule Mundi· Zenodo (CERN European Organi...· 0 citations
Reproducibility and replicability are our scientific and ethical mandate, and the foundation of trust in nursing science and research. Both rest on transparent documentation of how data were actually analyzed. The rising use of generative artificial intelligence (GenAI) in research intersects with this mandate, with th...
Gina M. Gehling, Lacey W. Heinsberg· Biological Research For Nurs...· 0 citations
The technical complexity and cost involved in electrical installation projects make them prone to risks. These risks, when not properly managed, often result to cost overruns. This study examined the effect of risk factors on cost overruns in electrical installation projects. The study focused on financial risk, techni...
Abayomi Adedayo Omonori, ADESINA NURUDEEN AJAO· International Journal of Env...· 0 citations
After selecting data of Montage Technology from 2021 to 2025, one can observe a growth process that is not neatly linear. The overall scale expanded over the five years, yet the differences between years were large. Based mainly on the company's annual reports, this paper continuously observes the changes in operating...
Manufacturing organizations record production, inventory, purchasing, and quality events in Enterprise Resource Planning (ERP) systems, yet these records are often fragmented, inconsistently timestamped, and reported with delay, which limits their usefulness for operational decisions. This paper proposes a modular fram...
Lalit Shirsath· Zenodo (CERN European Organi...· 0 citations
Is your company’s AI sovereign? A practical guide for CIOs on AI vendor lock-in, EU Data Act switching, cloud sovereignty, and testing your exit plan before you need it. The post Sovereignty Is the Option to Leave appeared first on GPT-Lab.
