Research Topics
We mainly focus on developing deep learning algorithms to solve many problems in the computer vision field. Including general tasks in computer vision such as classification or detection, we also study applicable work like generative models(GAN), depth estimation, medical imaging, noisy data, semantic segmentation, etc. Besides, there are many research on the fundamental structure of neural networks such as network minimization and architecture search in our laboratory. Our various research topics can be categorized into three sections which are Neural Network Architecture, Learning Methods, and Deep Learning Applications.
Deep Learning Applications
Our research topics in application of deep learning are focused on autonomous driving (depth estimation, object tracking, multi-sensor recognition), video recognition (video Turing test, video captioning), and defect detection (manufacturing, medical imaging).
Image Generation
Generating a fusion image with one's identity and one's pose
Point Cloud Recognition
Applying deep learning methods on 3D point cloud data acquired by LiDAR sensors
Video Recognition
Generating captions or answering questions on video
Object Tracking
Road environment recognition for autunomous driving
Defect Detection
Detecting abnormality in manufacturing process or medical images
Neural Network Architecture
Our research on neural network architectures aims to find the optimal structure for convolutional neural networks. Recent topics include designing new architectures, or new operations in the network, and automated architecture search.
Shape of Convolution Kernels
Learning the shape of convolution kernels in convolutional neural networks (Active Convolution)
Fast Convolution Operation
Deconstruction of 3x3 convolution operations into efficient 1x1 convolutions (Active Shift)
Network Architecture
Designing a new CNN architecture with gradually increasing feature dimension (PyramidNet)
Neural Architecture Search
Automated architecture search with deep learning
Learning Methods
We study various topics in learning methods to solve problems in data used for training, such as noisy labels or biases. We also aim to improve the utilization of neural networks in real life, by knowledge distillation, network pruning, and continual learning.
Noisy Labeled Data
Utilizing negative ground truth for robustness to noisy-labeled data (Negative Learning)
Biased Data
Learning useful information while not learning biases included in the data
Knowledge Distillation
Transferring knowledge from a network to another
Network Pruning
Minimizing neural networks by removing unnecessary parameters
Incremental Learning
Training a neural network with new data while keeping information learned from original data