September, 30th
9:00-9:05 - Introduction from CADL organizers
9:05-9:35 - "Scalable AI", Giuseppe Fiameni NVIDIA
9:35-10:20 - Short paper presentations
9:35 - 9:40 CycleBNN: Cyclic Precision Training in Binary Neural Networks
9:40 - 9:45 EPTQ: Enhanced Post-Training Quantization via Hessian-guided Network-wise Optimization
9:45 - 9:50 Generalized SAM: Efficient Fine-Tuning of SAM for Variable Input Image Sizes
9:50 - 9:55 ERF-NAS: Efficient Receptive Field-based Zero-Shot NAS for Object Detection
9:55 - 10:00 Optimizing Resource Consumption in Diffusion Models through Hallucination Early Detection
10:00 - 10:05 Mixed Non-linear Quantization for Vision Transformers
10:05 - 10:10 DailyMAE: Towards Pretraining Masked Autoencoders in One Day
10:10 - 10:15 Latent Distillation for Continual Object Detection at the Edge
10:15 - 10:20 MCUBench: A Benchmark of Tiny Object Detectors on MCUs
10:20-10:30 - Catch your short paper presenter
10:30-11:00 - Coffee break
11:00-12:15 - Full paper presentations (11 min. each + questions)
11:00 - 11:15 LightAvatar: Efficient Head Avatar as Dynamic Neural Light Field
11:15 - 11:30 CA3D: Convolutional-Attentional 3D Nets for Efficient Video Activity Recognition on the Edge
11:30 - 11:45 Famba-V: Fast Vision Mamba with Cross-Layer Token Fusion
11:45 - 12:00 Memory-Optimized Once-For-All network
12:00 - 12:15 Giving each task what it needs - leveraging structured sparsity for tailored multi-task learning
12:15-12:25 - CINECA
12:25 - 12:45 Open panel about future research directions
12:45-12:55 - Award ceremony and closing remarks
Accepted papers
Papers selected for full presentation
LightAvatar: Efficient Head Avatar as Dynamic Neural Light Field
Huan Wang (Northeastern University)*; Feitong Tan (Google); Ziqian Bai (Simon Fraser University); Yinda Zhang (Google); Shichen Liu (Google LLC); Qiangeng Xu (Google); Menglei Chai (Google); Anish J Prabhu (Google); Rohit Pandey (Google); Sean Fanello (Google); Zeng Huang (Snap Inc.); YUN FU (Northeastern University)
CA3D: Convolutional-Attentional 3D Nets for Efficient Video Activity Recognition on the Edge
Gabriele Lagani (CNR-ISTI)*; Fabrizio Falchi (CNR-ISTI); Claudio Gennaro (CNR-ISTI); Giuseppe Amato (CNR-ISTI)
Famba-V: Fast Vision Mamba with Cross-Layer Token Fusion
Hui Shen (The Ohio State University)*; Zhongwei Wan (The Ohio State University); Xin Wang (The Ohio State University); Mi Zhang (The Ohio State University)
Memory-Optimized Once-For-All network
Maxime Girard (Télécom Paris); Victor Quetu (Télécom Paris - Institut Polytechnique de Paris)*; Samuel Tardieu (Télécom Paris); Van-Tam Nguyen (Telecom Paris - Institut Polytechnique de Paris); Enzo Tartaglione (Télécom Paris - Institut Polytechnique de Paris)
Giving each task what it needs - leveraging structured sparsity for tailored multi-task learning
Richa Upadhyay (Luleå University of Technology)*; Ronald Phlypo (CNRS, Univ Grenoble Alpes, Grenoble INP); Rajkumar Saini (Luleå tekniska universitet, Luleå, Sweden); Marcus Liwicki (Luleå University of Technology)
Papers selected for short presentation
CycleBNN: Cyclic Precision Training in Binary Neural Networks
Federico Fontana (La Sapienza University of Rome)*; Romeo Lanzino (Sapienza University of Rome); Anxhelo Diko (La Sapienza University of Rome); Gian Luca Foresti (University of Udine, Italy); Luigi Cinque (University La Sapienza of Rome)
EPTQ: Enhanced Post-Training Quantization via Hessian-guided Network-wise Optimization
Ofir Gordon (Sony)*; Elad Cohen (Sony); Hai Victor Habi (Tel Aviv University); Arnon Netzer (Sony)
Generalized SAM: Efficient Fine-Tuning of SAM for Variable Input Image Sizes
Sota Kato (Meijo university); Hinako Mitsuoka (Meijo University)*; Kazuhiro Hotta (Meijo University)
ERF-NAS: Efficient Receptive Field-based Zero-Shot NAS for Object Detection
Xinyi Yu (Peking University)*; Runan Yin (Beijing University of Posts and Telecommunications); Zhihao Lin (Peking University); Yongtao Wang (Peking University)
Optimizing Resource Consumption in Diffusion Models through Hallucination Early Detection
Federico Betti (Univeristá di Trento)*; Lorenzo Baraldi (University of Pisa); Lorenzo Baraldi (University of Modena and Reggio Emilia); Rita Cucchiara (Università di Modena e Reggio Emilia); Nicu Sebe (University of Trento)
Mixed Non-linear Quantization for Vision Transformers
Gihwan Kim (Chungnam National University)*; Jemin Lee (Electronics and Telecommunications Research Institute); Sihyeong Park (Korea Electronics Technology Institute); Yongin Kwon (Electronics and Telecommunications Research Institute); Hyungshin Kim (Chungnam National University)
DailyMAE: Towards Pretraining Masked Autoencoders in One Day
Jiantao Wu (University of Surrey)*; Shentong Mo (Carnegie Mellon University); Sara Ahmed (University of surrey); Zhen-Hua Feng (University of Surrey); Josef Kittler (University of Surrey); Muhammad Awais (University of Surrey)
Latent Distillation for Continual Object Detection at the Edge
Francesco Pasti (University of Padova)*; Marina Ceccon (University of Padova); Davide Dalle Pezze (University of Padova); Francesco Paissan (Fondazione Bruno Kessler); Elisabetta Farella (Fondazione Bruno Kessler); Gian Antonio Susto (University of Padova); Nicola Bellotto (University of Padua)
MCUBench: A Benchmark of Tiny Object Detectors on MCUs
Sudhakar Sah (Deeplite Inc); Darshan C Ganji (Deeplite Inc.); Matteo Grimaldi (Deeplite): Ravish Kumar (Deeplite); Alexander Hoffman (McGill University); Honnesh Rohmetra (Deeplite, Inc.); Ehsan Saboori (Deeplite Inc.)*