Enterprise Organization of National Program of Excellence in Software

Two papers by the Visual Computing Laboratory (Guidance Professor: Jae-Pil Heo) have been approved for publication at the European Conference on Computer Vision 2022, a top-tier academic conference in the field of computer vision and artificial intelligence.

Thesis #1: "Tailoring Self-Supervision for Superimposed Learning"
(Master of Artificial Intelligence WonJun Moon, Ph.D. in Artificial Intelligence, Ji-Hwan Kim)
Thesis #2: "Difficulty-Aware Simulator for Open Set Recognition"
(Master of Artificial Intelligence WonJun Moon, Master of Artificial Intelligence Junho Park, Master of Artificial Intelligence Hyun-Seok Seong, Master of Artificial Intelligence, and Master of Artificial Intelligence Cheol-Ho Cho)

In the "Tailing Self-Supervision for Supervisory Learning" paper, we first pointed out the problems that may arise when the Self-supervision Task has been applied additionally without any special changes in the Supervisory Learning environment. When the Self-supervision Task is applied as an assistant to the Objective of Supervised Learning, we present three characteristics that Self-supervision Task should have and propose a new task called Localization Rotation that satisfies them. We find that the proposed method brings consistent performance improvements on several benchmarks that can test the robustness and generalization capabilities of the Deep Learning model. 

The paper "Difficulty-Aware Simulator for Open Set Recognition" presents a new method for simulating virtual samples for Open Set Recognition. The Open Set Recognition problem is a problem that identifies a new class of data that has not been experienced in learning and is an essential element of technology for applying artificial intelligence to the real world. Existing methods also generated and utilized virtual samples for model learning, but this paper confirmed that existing technologies are difficult to cope with Open Set samples of various difficulties. We propose a Difficulty-Aware Simulator framework that simulates Open Set samples of various difficulty levels. The proposed technique produced virtual samples by difficulty level from the standpoint of the classifier as intended. Using this, we achieved high performance in the Open Set Recognition field.

[Thesis #1 Information]

Tailoring Self-Supervision for Supervised Learning

WonJun Moon, Ji-Hwan Kim, and Jae-Pil Heo

European Conference on Computer Vision (ECCV), 2022

Abstract:

Recently, it is shown that deploying a proper self-supervision is a prospective way to enhance the performance of supervised learning. Yet, the benefits of self-supervision are not fully exploited as previous pretext tasks are specialized for unsupervised representation learning. To this end, we begin by presenting three desirable properties for such auxiliary tasks to assist the supervised objective. First, the tasks need to guide the model to learn rich features. Second, the transformations involved in the self-supervision should not significantly alter the training distribution. Third, the tasks are preferred to be light and generic for high applicability to prior arts. Subsequently, to show how existing pretext tasks can fulfill these and be tailored for supervised learning, we propose a simple auxiliary self-supervision task, predicting localizable rotation (LoRot). Our exhaustive experiments validate the merits of LoRot as a pretext task tailored for supervised learning in terms of robustness and generalization capability.

[Thesis #2 Information]

Difficulty-Aware Simulator for Open Set Recognition

WonJun Moon, Junho Park, Hyun Seok Seong, Cheol-Ho Cho, and Jae-Pil Heo

European Conference on Computer Vision (ECCV), 2022

Abstract:

Open set recognition (OSR) assumes unknown instances appear out of the blue at the inference time. The main challenge of OSR is that the response of models for unknowns is totally unpredictable. Furthermore, the diversity of open set makes it harder since instances have different difficulty levels. Therefore, we present a novel framework, DIfficulty-Aware Simulator (DIAS), that generates fakes with diverse difficulty levels to simulate the real world. We first investigate fakes from generative adversarial network (GAN) in the classifier's viewpoint and observe that these are not severely challenging. This leads us to define the criteria for difficulty by regarding samples generated with GANs having moderate-difficulty. To produce hard-difficulty examples, we introduce Copycat, imitating the behavior of the classifier. Furthermore, moderate- and easy-difficulty samples are also yielded by our modified GAN and Copycat, respectively. As a result, DIAS outperforms state-of-the-art methods with both metrics of AUROC and F-score.