About
Research & Development
Uncovering the Hidden Costs: Retrieval Biases in Multilingual AI for Corporate RAG
TLDR: This research paper investigates cross-lingual Retrieval-Augmented Generation (RAG) in Arabic-English corporate datasets, revealing that retrieval is a critical bottleneck, especially when user queries...
Research & Development
M4Doc: Boosting Machine Translation for Document Images
TLDR: M4Doc is a novel framework designed to improve Document Image Machine Translation (DIMT) by leveraging Multimodal Large Language Models (MLLMs). It uses a...
Research & Development
Unlocking Deeper Confidence in Language Models with Bayesian Discrete Diffusion
TLDR: A new research paper reveals that discrete diffusion language models inherently possess a Bayesian core, allowing them to achieve significantly lower perplexity (8.8...
Research & Development
TransformEEG: A New Deep Learning Model for Consistent Parkinson’s Disease Detection via EEG
TLDR: TransformEEG is a novel hybrid Convolutional-Transformer deep learning model designed to improve the generalizability and consistency of Parkinson's Disease detection using EEG data....
Research & Development
Unveiling Hadronic State Structures with Uncertainty-Aware Machine Learning
TLDR: This research introduces an uncertainty-aware machine learning framework to classify the pole structures of hadronic states from experimental data. By using an ensemble...
Research & Development
Achieving True Forgetting in AI Models: Introducing One-Point-Contraction Unlearning
TLDR: Machine unlearning often fails to truly remove data from AI models, leading to 'shallow forgetting' where internal information persists. Researchers propose 'One-Point-Contraction (OPC)'...