GitHub Activity
Explore my open-source contributions to AI research projects
Exploring the frontiers of artificial intelligence and machine learning
Selected research papers and publications in the field of AI
In this research, we explore efficient transformer architectures for languages with limited digital resources. Our approach reduces parameters by 60% while maintaining 95% accuracy compared to standard transformer models.
Through our novel pruning technique, we demonstrate that specialized transformer architectures can achieve state-of-the-art results for low-resource languages while requiring significantly less computational power and training data.
Traditional transformer models require massive amounts of data and computational resources, making them impractical for the majority of the world's languages. Our research addresses this gap by developing specialized architectures that can perform effectively with limited resources.
The proposed architecture incorporates language-specific adaptations and transfer learning techniques that allow for effective knowledge sharing between related languages. This approach has shown promising results across 15 different low-resource languages.
Read Full PaperOngoing and completed research in artificial intelligence
Exploring efficient transformer architectures for languages with limited digital resources. Our approach reduces parameters by 60% while maintaining 95% accuracy.
New differential privacy techniques for federated learning systems that improve privacy guarantees without sacrificing model performance on edge devices.
Implementing spiking neural networks on neuromorphic hardware for energy-efficient AI at the edge. Achieved 40x power reduction compared to GPUs.
Novel approach to combining visual, textual, and sensor data in reinforcement learning models for more robust decision-making in complex environments.
Leveraging graph neural networks to capture complex item-user relationships in recommender systems, improving recommendation accuracy by 18%.
Techniques for transferring language models across linguistically diverse languages using minimal examples, enabling NLP capabilities for under-resourced languages.
Explore my open-source contributions to AI research projects