김지섭 교수

Generalized zero-shot learning (GZSL) is a technique to train a deep learning model to identify unseen classes using the image attribute. In this paper, we put forth a new GZSL technique exploiting Vision Transformer (ViT) to maximize the attribute-related information contained in the image feature. In ViT, the entire image region is processed without the degradation of the image resolution and the local image information is preserved in patch features. To fully enjoy the benefits of ViT, we exp

Abstract Before the global financial crisis, the proportion of households defaulting on the mortgage while remaining current on the unsecured loan was almost the same as the proportion of households current on the mortgage but defaulting on the unsecured loan. After the crisis, the former ratio became higher than the latter. By using a heterogeneous agent model with the mortgage and the unsecured loan, I examine how the order of defaults changed before and after the crisis. I then analyze the im

In this paper, we suggest the testing method for industrial level cyber-physical system and its automation under the various environmental, restrictive condition. There are other testing methods, such as Software in the Loop Test, Hardware in the Loop Test, Field Test. But these methods have limitations of high cost, not enough testing and hard to test because the complexity of the situation is going up. The systems that work near human, like drones, autonomous cars, or used in extreme environme

Face-swapping models have been drawing attention for their compelling generation quality, but their complex architectures and loss functions often require careful tuning for successful training. We propose a new face-swapping model called `Smooth-Swap', which excludes complex handcrafted designs and allows fast and stable training. The main idea of Smooth-Swap is to build smooth identity embedding that can provide stable gradients for identity change. Unlike the one used in previous models train

We present an encoder-powered generative adversarial network (EncGAN) that is able to learn both the multi-manifold structure and the abstract features of data. Unlike the conventional decoder-based GANs, EncGAN uses an encoder to model the manifold structure and invert the encoder to generate data. This unique scheme enables the proposed model to exclude discrete features from the smooth structure modeling and learn multi-manifold data without being hindered by the disconnections. Also, as EncG

This paper analyzes how and why household debt distribution by the householder age has changed over the past decade both in Korea and the US. Data shows that the proportion of household debt held by younger households has decreased, while that held by older households has increased. Empirical analysis shows that a change in the demographic distribution of householders is the main driving force that has shifted the household debt distribution. Given that demographic aging is an inevitable trend,

This paper analyses structural changes in demographic, social, and economic conditions in North Korea and draws policy implications on housing supply and residential environment. Demographic and social structures in North Korea, such as population aging, low fertility rate, and increases in nuclear families, are changing, just as many developed countries have experienced. At the same time, there is a high demand for house and infrastructure redevelopment. Meanwhile, there are significant differe

Exploiting the deep generative model's remarkable ability of learning the data-manifold structure, some recent researches proposed a geometric data interpolation method based on the geodesic curves on the learned data-manifold. However, this interpolation method often gives poor results due to a topological difference between the model and the dataset. The model defines a family of simply-connected manifolds, whereas the dataset generally contains disconnected regions or holes that make them non

This paper analyzes why the household debt held by seniors in Korea is highly vulnerable, compared to the US and major European countries. Based on household-level micro data, seniors in Korea have lower income, lower income stability, and fewer financial assets than those in the US and European countries. In addition, the macro-financial environment over the last decade in Korea promoted debt accumulation. Hence, if Korean economy is hit by adverse macro-financial shocks, such as a sudden incre