Post: Real-Time Semantic Segmentation
#ENet model implementation In this blog I will present an overview on image segmentation and the reproduction of ENet paper.
I hold a PhD in Machine Learning from Bar Ilan University, where I worked at ChechikLab under the supervision of Prof. Gal Chechik, head of the Computational Neurobiology Lab & Nvidia Research Group IL. My research focuses on multi-task learning, learning in weight-space methods, and generative AI for speech.
Currently, I work as Data Science Tech Lead at aiOla, where I lead the technical design and architecture of data science projects, driving research through publications in top-tier conferences.
Recent Updates
- Whisper in Medusa's Ear: Multi-head Efficient Decoding for Transformer-based ASR accepted to ICASSP 2025!
- WhisperNER: Unified Open Named Entity and Speech Recognition accepted to ASRU 2025!
- Target Speaker Extraction with Flow Matching accepted to InterSpeech 2025!
- UmbraTTS: Adapting Text-to-Speech to Environmental Contexts with Flow Matching accepted to ICML WMLA 2025!
- Improved Generalization of Weight Space Networks via Augmentations accepted to ICML 2024!
- Keyword-guided Adaptation of Automatic Speech Recognition accepted to InterSpeech 2024!
- Equivariant Deep Weight Space Alignment accepted to ICML 2024!
- LipVoicer: Generating Speech from Silent Videos Guided by Lip Reading accepted to ICLR 2024!
- FedSelect: Personalized Federated Learning with Customized Selection of Parameters for Fine-Tuning accepted to CVPR 2024!
- Open-Vocabulary Keyword-Spotting with Adaptive Instance Normalization accepted to ICASSP 2024!