Research

1. Statistical Mechanics and Complex Systems

Figure 1. (a) Schematic illustration of protein-protein interction (PPI) network under global suppression process. As total node number N increases with time, the smallest clusters of gN nodes belong to set R and the rest belong to set Rc. At each time step, a new node is added and duplicates link of a target node with probability 1-q. Subsequently, the new node is linked to a randomly selected node among set R with probability p. (b) Phase diagram for PPI network under global suppression process for given g and p. We can control hyperparamters but there always exist three phases.

I have investigated the percolation transitions in growing systems resembling the real world including protein-protein interaction network (Figure 1(a)), co-authorship network, and others. First, we develop the numerical and theoretical model to investigate when a giant cluster emerges. We then derive the governing equation of the cluster size distribution. By solving this master equation in steady state, various physical variables can be calculated. We finally get the phase diagram where there exist three phases for given hyperparameters as shown in Figure 1(b).

Publications:

[11] S. M. Oh and B. Vidakovic, "Wavelet-Based Coarse Graining for Percolation Criticality from a Single System Size", under review at Chaos (2025). [DOI in preparation]

[10] S. M. Oh, Y. Lee, and B. Kahng, "Emergent Properties in Social Networks with Higher-Order Interactions", under review at Chaos Solitons, & Fractals (2025). [DOI in preparation]

[9] S. M. Oh, K. Choi, and B. Kahng, "Machine Learning Approach to Percolation Transitions: Global Information", Journal of Statistical Mechanics 2023, 083210 (2023). [DOI link] [IF: 2.2]

[8] Y. Jeong, S. M. Oh, and Y. S. Cho, "Discontinuous Emergence of a Giant Cluster in Assortative Scale-Free Networks", Journal of the Korean Physical Society 81, 608 (2022). [DOI link] [IF: 0.4]

[7] S. M. Oh, Y. Lee, J. Lee, and B. Kahng, "Emergence of Betti Numbers in Growing Simplicial Complexes: Analytical Solutions", Journal of Statistical Mechanics 2021, 083218 (2021). [DOI link] [IF: 2.2]

[6] J. Lee, Y. Lee, S. M. Oh, and B. Kahng, "Betweenness Centrality of Teams in Social Networks", Chaos 31, 061108 (2021). [DOI link] [IF: 2.7]

[5] S. M. Oh, S.-W. Son and B. Kahng, "Percolation Transitions in Growing Networks Under Achlioptas Processes: Analytic Solutions", Chaos, Solitons & Fractals 146, 110889 (2021). [DOI link] [IF: 5.3]

[4] Y. Lee, J. Lee, S. M. Oh, D. Lee, and B. Kahng, "Homological Percolation Transitions in Growing Simplicial Complexes", Chaos 31, 041102 (2021). [DOI link] [IF: 2.7]

[3] S. M. Oh, S.-W. Son and B. Kahng, "Discontinuous Percolation Transitions in Growing Networks", Journal of Statistical Mechanics 2019, 083502 (2019). [DOI link] [IF: 2.2]

[2] S. M. Oh, S.-W. Son and B. Kahng, "Suppression Effect on the Berezinskii-Kosterlitz-Thouless Transition in Growing Networks", Physical Review E 98, 060301(R) (2018). [DOI link] [IF: 2.2]

[1] S. M. Oh, S.-W. Son and B. Kahng, "Explosive Percolation Transitions in Growing Networks", Physical Review E 93, 032316 (2016). [DOI link] [IF: 2.2]

Figure 2. (a) Plot of geostationary AMVs in Typhoon Nepartak at 00:00 UTC on 7 July 2016. Colors represent pressure altitude of AMVs. The background plot is the CH13 (10.2 m) brightness temperature of AHI. (b) Plot of Comparison of AMV heights and Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) level 1 lidar products for 05:40 UTC 21 July 2016. CALIOP 532 attenuated backscatter data is used.

I have developed an atmospheric motion vector (AMV) retrieval algorithm using the geostationary satellite images. Our AMV algorithm consists of four main steps such as target selection, height assignment, tracking, and quality control. We used the visible and infrared channels of Himawari-8/AHI for selecting and tracking target, the numerical weather prediction (NWP) data for analyzing the inversion layer, and the radiative transfer model for assigning height. The retrieved AMVs can help to understand and predict the atmosphere around severe weather as shown in Figure 2(a). Figure 2(b) confirms that our AMV heights are consistent with MTG AMV results from EUMETSAT and CALIPSO CALIOP profiles regardless of height assignment methods.

Publications:

[3] H.-A. Kim, S.-R. Chung, S. M. Oh, B.-I. Lee, and I.-C, Shin, "The Impact of Spatio-temporal Resolution of GEO-KOMPSAT-2A Rapid Scan Imagery on the Retrieval of Mesoscale Atmospheric Motion Vector", Korean Journal of Remote Sensing 37, 885 (2021). [DOI link] [IF: 0.9]

[2] D. Santek, ..., S. M. Oh (12th/16), et al., "2018 Atmospheric Motion Vector (AMV) Intercomparison Study", Remote Sensing 11, 2240 (2019). [DOI link] [IF: 4.2]

[1] S. M. Oh, R. Borde, M. Carranza, and I.-C, Shin, "Development and Intercomparison Study of Atmospheric Motion Vector Retrieval Algorithm for GEO-KOMPSAT-2A", Remote Sensing 11, 2054 (2019). [DOI link] [IF: 4.2]

2. Quantum Network Science

In preparation

Publications:

[2] S. M. Oh, S. Marano, H. Shin, A. Conti and M. Z. Win, "Noise-Resilient Entanglement Percolation in Quantum Networks via Higher-Order Interactions", under review at PRX Quantum (2025). [DOI in preparation]

[1] S. M. Oh, H. Shin, S. Marano, A. Conti, and M. Z. Win, "Entanglement Percolation in Noisy Quantum Networks", in 2024 IEEE International Conference on Quantum Communications, Networking, and Computing, (2024), pp. 143-149. [DOI link] [Best Paper Award]

3. Machine Learning

Figure 3. Architectures of (a) convolutional autoencoder and (b) convolutional neural network we used.

I'm trying to understand various physical phenomena based on machine learning (ML) approach. First of all, we use unsupervised and supervised learning to investigate percolation properties in networks as shown in Figure 3. We are preparing the corresponding paper. Our preliminary results confirm that ML approach is successfully applicable to investigate the percolation transitions in networks.

Publications:

[3] S. M. Oh, Y. Li, and J. W. Chong, "Wavelet Convolutional Neural Network for Low-Resolution Brain MRI Images", in 2025 IEEE International Symposium on Biomedical Imaging, accepted (2025). [DOI in preparation]

[2] S. M. Oh, K. Choi, and B. Kahng, "Machine Learning Approach to Percolation Transitions: Global Information", Journal of Statistical Mechanics 2023, 083210 (2023). [DOI link] [IF: 2.2]

[1] J. W. Song, K. Choi, S. M. Oh, and B. Kahng, "Exploring Nonlinear Dynamics and Network Structures in Kuramoto Systems Using Machine Learning Approaches", Chaos 33, 073148 (2023). [DOI link] [IF: 2.7]

Page updated

Google Sites

Report abuse