Doctoral thesis
Open access

Learning to compress and search visual data in large-scale systems

ContributorsFerdowsi, Sohraborcid
Defense date2018-12-11

The problem of high-dimensional and large-scale representation of visual data is addressed from an unsupervised learning perspective, where an emphasis is put on discrete representations. The algorithmic infrastructure is developed based on the synthesis and analysis prior models whose rate-distortion properties are carefully optimized. These are then extended to multi-layers, namely the RRQ and the ML-STC frameworks, where the latter is further evolved as a powerful deep neural network architecture with fast and sample-efficient training. For these frameworks, three important applications are considered. First, large-scale similarity search in retrieval systems is addressed, where a double-stage solution is proposed leading to faster query times and smaller storage. Second, the problem of learned image compression is targeted, where the proposed models can capture more redundancies within images than the conventional compression codecs. Finally, the proposed algorithms are used to solve ill-posed inverse problems with promising results in image denoising and compressive sensing.

  • Unsupervised learning
  • Representation learning
  • Learned compression
  • Similarity search
  • Approximate nearest neighbor search
  • Rate-distortion theory
  • Ill-posed inverse problems
  • Image processing
Citation (ISO format)
FERDOWSI, Sohrab. Learning to compress and search visual data in large-scale systems. 2018. doi: 10.13097/archive-ouverte/unige:114990
Main files (1)

Technical informations

Creation03/12/2019 2:14:00 PM
First validation03/12/2019 2:14:00 PM
Update time02/08/2024 12:40:24 PM
Status update02/08/2024 12:40:24 PM
Last indexation05/03/2024 7:44:25 AM
All rights reserved by Archive ouverte UNIGE and the University of GenevaunigeBlack