TRACK 2: Deep Learning and Optimization


Deep Learning and Optimization encompass a wide array of techniques employed in pattern recognition and machine learning (PRML). Deep learning, a subset of machine learning, employs artificial neural networks with multiple layers (deep architectures) to understand and interpret data patterns. Neural generative models, including autoencoders and Generative Adversarial Networks (GANs), enable the creation of new data instances resembling the training set. Autoencoders compress data while GANs generate new instances through a duelling network process. These tools further aid in scene analysis and understanding, enabling machines to make sense of complex visual environments. The field also includes multi-view learning, transfer learning, low-shot, semi-, and unsupervised learning methods. Multi-view learning aims to improve the model’s performance by understanding things better from various perspectives. Transfer learning enhances learning efficiency and performance when a pre-trained model is used on a new problem. Low-shot learning deals with scarce data scenarios, while unsupervised learning identifies hidden patterns in unlabelled data. Finally, motion and tracking engage these tools to predict and track dynamic object movement, vital for applications such as autonomous driving or video surveillance. All these methodologies are critical for pushing the frontiers of PRML and its applications.

Track Chairs:
     Prof. Quanxue Gao, Xidian University, China
     Prof. Feiping Nie, Northwestern Polytechnical University, China

Track Program Chairs:
     Assoc. Prof. Qianqian Wang, Xidian University, China
     Assoc. Prof. Ming Yang, the University of Evansville, USA
     Assoc. Prof. Deyan Xie, Qingdao Agricultural University, China

Track Technical Committee:
     Dr. Xia Wei, Xidian University, China
     Asst. Prof. Danyang Wu, Xi'an Jiaotong University, China
     Asst. Prof. Zheng Wang, Northwestern Polytechnical University, China
     Dr. Wenxuan Tu, National University of Defense Technology, China
     Dr. Canyu Zhang, Northwestern Polytechnical University, China
     Assoc. Prof. Han Zhang, Northwestern Polytechnical University, China

Topics of interest include, but are not limited to:

    ◆ Neural Generative Models, Autoencoders, GANs
    ◆ Optimization and Learning Methods
    ◆ Representation Learning and Deep Learning
    ◆ Scene Analysis and Understanding
    ◆ Transfer Learning, Low-Shot, Semi- and Unsupervised Learning
    ◆ Motion and Tracking

Submission Guidelines

Please submit your manuscript via Electronic Submission System (account is needed). (Please choose the track number when you make the submission.)

Important Dates

    ◆ Submission of Full Papers: February 20, 2024
    ◆ Notification of Review Result of Papers from Track: March 20, 2024
    ◆ Registration Deadline: March 30, 2024