Výsledky vyhledávání

Report

Interpreting token compositionality in LLMs: A robustness analysis

Autor: Aljaafari, Nura, Carvalho, Danilo S., Freitas, André

Understanding the internal mechanisms of large language models (LLMs) is integral to enhancing their reliability, interpretability, and inference processes. We present Constituent-Aware Pooling (CAP), a methodology designed to analyse how LLMs proces

Externí odkaz: http://arxiv.org/abs/2410.12924

Zobrazit plný text záznamu

Report

Structuring a Training Strategy to Robustify Perception Models with Realistic Image Augmentations

Autor: Hammam, Ahmed, Sreedhar, Bharathwaj Krishnaswami, Kawa, Nura, Patzelt, Tim, De Candido, Oliver

Advancing Machine Learning (ML)-based perception models for autonomous systems necessitates addressing weak spots within the models, particularly in challenging Operational Design Domains (ODDs). These are environmental operating conditions of an aut

Externí odkaz: http://arxiv.org/abs/2408.17311

Zobrazit plný text záznamu

Report

The Mechanics of Conceptual Interpretation in GPT Models: Interpretative Insights

Autor: Aljaafari, Nura, Carvalho, Danilo S., Freitas, André

Locating and editing knowledge in large language models (LLMs) is crucial for enhancing their accuracy, safety, and inference rationale. We introduce ``concept editing'', an innovative variation of knowledge editing that uncovers conceptualisation me

Externí odkaz: http://arxiv.org/abs/2408.11827

Zobrazit plný text záznamu

Report

Seeds of Stereotypes: A Large-Scale Textual Analysis of Race and Gender Associations with Diseases in Online Sources

Autor: Hansen, Lasse Hyldig, Andersen, Nikolaj, Gallifant, Jack, McCoy, Liam G., Stone, James K, Izath, Nura, Aguirre-Jerez, Marcela, Bitterman, Danielle S, Gichoya, Judy, Celi, Leo Anthony

Background Advancements in Large Language Models (LLMs) hold transformative potential in healthcare, however, recent work has raised concern about the tendency of these models to produce outputs that display racial or gender biases. Although training

Externí odkaz: http://arxiv.org/abs/2405.05049

Zobrazit plný text záznamu

Report

Genetic Algorithm-Based Dynamic Backdoor Attack on Federated Learning-Based Network Traffic Classification

Autor: Nazzal, Mahmoud, Aljaafari, Nura, Sawalmeh, Ahmed, Khreishah, Abdallah, Anan, Muhammad, Algosaibi, Abdulelah, Alnaeem, Mohammed, Aldalbahi, Adel, Alhumam, Abdulaziz, Vizcarra, Conrado P., Alhamed, Shadan

Federated learning enables multiple clients to collaboratively contribute to the learning of a global model orchestrated by a central server. This learning scheme promotes clients' data privacy and requires reduced communication overheads. In an appl

Externí odkaz: http://arxiv.org/abs/2310.06855

Zobrazit plný text záznamu

Report

A Conditional Generative Chatbot using Transformer Model

Autor: Esfandiari, Nura, Kiani, Kourosh, Rastgoo, Razieh

A Chatbot serves as a communication tool between a human user and a machine to achieve an appropriate answer based on the human input. In more recent approaches, a combination of Natural Language Processing and sequential models are used to build a g

Externí odkaz: http://arxiv.org/abs/2306.02074

Zobrazit plný text záznamu

Report

An Efficient Transfer Learning-based Approach for Apple Leaf Disease Classification

Autor: Ashmafee, Md. Hamjajul, Ahmed, Tasnim, Ahmed, Sabbir, Hasan, Md. Bakhtiar, Jahan, Mst Nura, Rahman, A. B. M. Ashikur

Correct identification and categorization of plant diseases are crucial for ensuring the safety of the global food supply and the overall financial success of stakeholders. In this regard, a wide range of solutions has been made available by introduc

Externí odkaz: http://arxiv.org/abs/2304.06520

Zobrazit plný text záznamu

Vyhledávací nástroje:

Upřesnit hledání