Zobrazeno 1 - 10
of 18
pro vyhledávání: '"Chaudhari Harsh"'
Large language models (LLMs) are susceptible to memorizing training data, raising concerns due to the potential extraction of sensitive information. Current methods to measure memorization rates of LLMs, primarily discoverable extraction (Carlini et
Externí odkaz:
http://arxiv.org/abs/2410.19482
Privacy-preserving machine learning (PPML) enables multiple data owners to contribute their data privately to a set of servers that run a secure multi-party computation (MPC) protocol to train a joint ML model. In these protocols, the input data rema
Externí odkaz:
http://arxiv.org/abs/2409.15126
Autor:
Chaudhari, Harsh, Severi, Giorgio, Abascal, John, Jagielski, Matthew, Choquette-Choo, Christopher A., Nasr, Milad, Nita-Rotaru, Cristina, Oprea, Alina
Retrieval Augmented Generation (RAG) expands the capabilities of modern large language models (LLMs), by anchoring, adapting, and personalizing their responses to the most relevant knowledge sources. It is particularly useful in chatbot applications,
Externí odkaz:
http://arxiv.org/abs/2405.20485
Publikováno v:
SICOT-J, Vol 8, p 46 (2022)
Introduction: Surgical treatment is usually recommended for the acute unstable acromioclavicular joint (ACJ) dislocations. Among the wide variety of different surgical techniques, the Double Endobutton and the Nottingham Surgilig technique are two of
Externí odkaz:
https://doaj.org/article/3846bafb74a444ca828175d1100f1a8b
This work introduces the L3Cube-MahaSocialNER dataset, the first and largest social media dataset specifically designed for Named Entity Recognition (NER) in the Marathi language. The dataset comprises 18,000 manually labeled sentences covering eight
Externí odkaz:
http://arxiv.org/abs/2401.00170
Autor:
Chaudhari, Harsh, Patil, Anuja, Lavekar, Dhanashree, Khairnar, Pranav, Joshi, Raviraj, Pande, Sachin
Named Entity Recognition (NER) systems play a vital role in NLP applications such as machine translation, summarization, and question-answering. These systems identify named entities, which encompass real-world concepts like locations, persons, and o
Externí odkaz:
http://arxiv.org/abs/2312.01306
The integration of machine learning (ML) in numerous critical applications introduces a range of privacy concerns for individuals who provide their datasets for model training. One such privacy risk is Membership Inference (MI), in which an attacker
Externí odkaz:
http://arxiv.org/abs/2310.03838
Autor:
Chaudhari, Harsh, Abascal, John, Oprea, Alina, Jagielski, Matthew, Tramèr, Florian, Ullman, Jonathan
Property inference attacks allow an adversary to extract global properties of the training dataset from a machine learning model. Such attacks have privacy implications for data owners sharing their datasets to train machine learning models. Several
Externí odkaz:
http://arxiv.org/abs/2208.12348
Secure multiparty computation (MPC) has been proposed to allow multiple mutually distrustful data owners to jointly train machine learning (ML) models on their combined data. However, by design, MPC protocols faithfully compute the training functiona
Externí odkaz:
http://arxiv.org/abs/2205.09986
Machine learning has started to be deployed in fields such as healthcare and finance, which propelled the need for and growth of privacy-preserving machine learning (PPML). We propose an actively secure four-party protocol (4PC), and a framework for
Externí odkaz:
http://arxiv.org/abs/1912.02631