Beyond fine-tuning: LoRA modules boost near-OOD detection and LLM security

À propos
Confidentialité
Mentions légales

Graph Chatbot

Publications associées (29)

Nothing New Under the Sun: Novelty Constructs and Measures in Social Studies

Davide Bavato

The concept of novelty is central to questions of creativity, innovation, and discovery. Despite the prominence in scientific inquiry and everyday discourse, there is a chronic ambiguity over its meaning and a surprising variety of empirical measures, whic ...

Emerald Publishing Limited2022

A Compressive Sensing Framework for EMI Source Localization Using a Metalens Structure: Localization Beyond the Diffraction Limit

Farhad Rachidi-Haeri, Marcos Rubinstein, Mohammad Azadifar, Hamidreza Karami

In this article, we propose a novel approach for the localization of electromagnetic interference (EMI) sources based on the concept of compressive sensing and metalenses. We show that using the spatial and frequency dispersion offered by the metalens, we ...

2022

Examining African contributions to global health: Reflections on knowledge circulation and innovation

Jérôme Chenal, Vitor Pessoa Colombo, Julia Tischler, Akuto Akpedze Konou, Jürg Utzinger

This article unites different disciplinary debates on ‘southern innovation’, ‘theory from the South’, and ‘decolonisation of knowledge’ in order to discuss existing understandings around the role of Africa in the production of health-related knowledge, pub ...

2022

Quality criteria for multi-domain studies in the indoor environment: Critical review towards research guidelines and recommendations

Mandana Sarey Khanie, Giorgia Chinazzo, Dolaana Khovalyg, Verena Marie Barthelmes, Matteo Favero

The perception, physiology, behavior, and performance of building occupants are influenced by multi-domain exposures: the simultaneous presence of multiple environmental stimuli, i.e., visual, thermal, acoustic, and air quality. Despite being extensive, th ...

PERGAMON-ELSEVIER SCIENCE LTD2022

Mechanical intelligence for learning embodied sensor-object relationships

Ahalya Prabhakar

Intelligence involves processing sensory experiences into representations useful for prediction. Understanding sensory experiences and building these contextual representations without prior knowledge of sensor models and environment is a challenging unsup ...

Nature Portfolio2022

Spatial Isolation Implies Zero Knowledge Even in a Quantum World

Alessandro Chiesa

Zero knowledge plays a central role in cryptography and complexity. The seminal work of Ben-Or et al. (STOC 1988) shows that zero knowledge can be achieved unconditionally for any language in NEXP, as long as one is willing to make a suitable physical assu ...

ASSOC COMPUTING MACHINERY2022

Interpreting Language Models Through Knowledge Graph Extraction

Martin Jaggi, Angeliki Romanou, Vinitra Swamy

Transformer-based language models trained on large text corpora have enjoyed immense popularity in the natural language processing community and are commonly used as a starting point for downstream tasks. While these models are undeniably useful, it is a c ...

2021

Product of experts for robot learning from demonstration

Emmanuel Pignat

Adaptability and ease of programming are key features necessary for a wider spread of robotics in factories and everyday assistance. Learning from demonstration (LfD) is an approach to address this problem. It aims to develop algorithms and interfaces such ...

EPFL2021

Unbiased Monte Carlo cluster updates with autoregressive neural networks

Giuseppe Carleo, Dian Wu, Riccardo Rossi

Efficient sampling of complex high-dimensional probability distributions is a central task in computational science. Machine learning methods like autoregressive neural networks, used with Markov chain Monte Carlo sampling, provide good approximations to s ...

AMER PHYSICAL SOC2021

Page 2 sur 2