Eunsom Jeon

Dr. Eunsom Jeon

orcid

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
29 December 2021 in IEEE Internet of Things Journal
Verified
Journal Article
Closed access
EJEunsom Jeon
Eunsom Jeon
ASAnirudh Som
Anirudh Som
ASAnkita Shukla
Ankita Shukla
KHKristina Hasanaj
Kristina Hasanaj
MBMatthew Buman
Matthew Buman
PTPavan Turaga
Pavan Turaga
Deep neural networks are parametrized by several thousands or millions of parameters and have shown tremendous success in many classification problems. However, the large number of parameters makes it difficult to integrate these models into edge devices, such as smartphones and wearable devices. To address this problem, knowledge distillation (KD) has been widely employed, which uses a pretrained high-capacity network to train a much smaller network, suitable for edge devices. In this article, for the first time, we study the applicability and challenges of using KD for time-series data for wearable devices. Successful application of KD requires specific choices of data augmentation methods during training. However, it is not yet known if there exists a coherent strategy for choosing an augmentation approach during KD. In this article, we report the results of a detailed study that compares and contrasts various common choices and some hybrid data augmentation strategies in KD-based human activity analysis. Research in this area is often limited as there are not many comprehensive databases available in the public domain from wearable devices. Our study considers databases from small-scale publicly available to one derived from a large-scale interventional study into human activity and sedentary behavior. We find that the choice of data augmentation techniques during KD has a variable level of impact on end performance, and find that the optimal network choice as well as data augmentation strategies are specific to a data set at hand. However, we also conclude with a general set of recommendations that can provide a strong baseline performance across databases.

20 December 2023 in Engineering Applications of Artificial Intelligence
Verified
Journal Article
Closed access
EJEunsom Jeon
Eunsom Jeon
HCHongjun Choi
Hongjun Choi
ASAnkita Shukla
Ankita Shukla
YWYuan Wang
Yuan Wang
HLHyunglae Lee
Hyunglae Lee
PTPavan Turaga
Pavan Turaga

11 June 2024 in IEEE Internet of Things Journal
Verified
Journal Article
Closed access
EJEunsom Jeon
Eunsom Jeon
MBMatthew Buman
Matthew Buman
PTPavan Turaga
Pavan Turaga
In applications involving analysis of the wearable sensor data, machine learning techniques that use features from the topological data analysis (TDA) have demonstrated remarkable performance. Persistence images (PIs) generated through TDA prove effective in capturing robust features, especially to signal perturbations, thus complementing classical time-series features. Despite its promising performance, utilizing TDA to create PI entails significant computational resources and time, posing challenges for applications on small devices. Knowledge distillation (KD) emerges as a solution to address these challenges, as it can produce a compact model. Using multiple teachers one trained with raw time-series and another with topological features, is a viable approach to distill a single compact student model. In such a case, the two teachers will have different statistical characteristics and need some form of feature harmonization. To tackle these issues, we propose uncertainty-aware topological persistence guided KD. This approach involves separating common and distinct components between the teachers and applying varying weights to control their effects. To enhance the knowledge provided to a student, uncertain features from the teachers are rectified using the uncertainty scores. We leverage feature similarities to offer more valuable information and employ relationships computed based on the orthogonal properties to prevent excessive feature transformation. Ultimately, our method yields a robust single student that operates solely on the time-series data at test time. We validate the effectiveness of the proposed approach through the empirical evaluations across various combinations of the models and data sets, demonstrating its robustness and efficacy in different scenarios. The proposed method enhances the classification performance of a student model by approximately 4.3% compared to a model learned from scratch on GENEActiv.

9 June 2025 in IEEE Internet of Things Journal
Verified
Journal Article
Closed access
EJEunsom Jeon
Eunsom Jeon
SMSinjini Mitra
Sinjini Mitra
JLJisoo Lee
Jisoo Lee
OSOmik M. Save
Omik M. Save
ASAnkita Shukla
Ankita Shukla
PTPavan Turaga
Pavan Turaga
Human gait analysis with wearable sensors has been widely used in various applications, such as daily life healthcare, rehabilitation, physical therapy, and clinical diagnostics and monitoring. In particular, ground reaction force (GRF) provides critical information about how the body interacts with the ground during locomotion. Although instrumented treadmills have been widely used as the gold standard for measuring GRF during walking, their lack of portability and high cost make them impractical for many applications. As an alternative, low-cost, portable, wearable insole sensors have been utilized to measure GRF; however, these sensors are susceptible to noise and disturbance and are less accurate than treadmill measurements. Deep learning has shown potential in addressing these issues, but such methods are computationally expensive and often require extensive computing resources, limiting their feasibility for real-time and portable systems. To address these challenges, we propose a Time-aware Knowledge Distillation framework for GRF estimation from insole sensor data. This framework leverages similarity and temporal features within a mini-batch during the knowledge distillation process, effectively capturing the complementary relationships between features and the sequential properties of the target and input data. The performance of the lightweight models distilled through this framework was evaluated by comparing GRF estimations from insole sensor data against measurements from an instrumented treadmill. Various teacher-student model architectures and learning strategies were evaluated across multiple performance metrics using data collected at different walking speeds. Empirical results demonstrated that Time-aware Knowledge Distillation outperforms current baselines in GRF estimation from wearable sensor data. Moreover, our method significantly reduces the number of training parameters needed for GRF estimation, offering a data-and resource-efficient solution for human gait analysis while achieving excellent accuracy and model reliability.

3 November 2023 in IEEE Transactions on Instrumentation and Measurement
Verified
Journal Article
Closed access
EJEunsom Jeon
Eunsom Jeon
HCHongjun Choi
Hongjun Choi
ASAnkita Shukla
Ankita Shukla
YWYuan Wang
Yuan Wang
MBMatthew Buman
Matthew Buman
PTPavan Turaga
Pavan Turaga
Wearable sensor data analysis with persistence features generated by topological data analysis (TDA) has achieved great success in various applications, and however, it suffers from large computational and time resources for extracting topological features. In this article, our approach utilizes knowledge distillation (KD) that involves the use of multiple teacher networks trained with the raw time series and persistence images (PIs) generated by TDA. However, direct transfer of knowledge from the teacher models utilizing different characteristics as inputs to the student model results in a knowledge gap and limited performance. To address this problem, we introduce a robust framework that integrates multimodal features from two different teachers and enables a student to learn desirable knowledge effectively. To account for statistical differences in multimodalities, an entropy-based constrained adaptive weighting mechanism is leveraged to automatically balance the effects of teachers and encourage the student model to adequately adopt the knowledge from two teachers. To assimilate dissimilar structural information generated by different style models for distillation, batch and channel similarities within a mini-batch are used. We demonstrate the effectiveness of the proposed method on wearable sensor data.

2 July 2025 in IEEE Internet of Things Journal
Verified
Journal Article
Closed access
SJSeong In Jeong
Seong In Jeong
MJMin Su Jeong
Min Su Jeong
EJEunsom Jeon
Eunsom Jeon
KPKang Ryoung Park
Kang Ryoung Park
In autonomous driving with Internet of Things (IoT) devices, real-time road perception and semantic segmentation are essential for intelligent transportation systems. However, deploying models on IoT devices is challenging due to their limited computational power, memory, and energy availability. Additionally, limited field of view (FoV) occurs frequently in real-world scenarios, such as constrained camera angles or occlusion by large objects, leading to significant degradations in segmentation performance. To address these challenges, we propose Dynamic Class Relation Knowledge Distillation (DCR-KD), a framework that generates lightweight models by transferring knowledge from high-performance teacher models. Central to our method is the Limited FoV Edge Module (LFEM), which extracts edge-aware features of the teacher to refine the learning of the student. LFEM is designed to capture edge-based features in regions with limited FoV, effectively representing critical object boundaries. A channel attention mechanism further enhances semantic features, allowing the student to focus on key information in constrained visual contexts. A dynamic class relation map captures global semantic relationships among classes, enriching the scene understanding of the student. The final student model is independent of the teacher during inference, enabling efficient deployment in resource-constrained environments. Extensive evaluations demonstrate the effectiveness of DCR-KD, including segmentation performance and feature visualizations. Our method bridges the performance gap between resource-intensive teacher models and efficient student models, providing a practical solution for IoT-based real-time road perception, particularly under limited FoV conditions. The code and pre-trained models are publicly available at.