Minji Lee

Sumin Han, Youngjun Park, Minji Lee et al.

Traffic prediction is one of the key elements to ensure the safety and convenience of citizens. Existing traffic prediction models primarily focus on deep learning architectures to capture spatial and temporal correlation. They often overlook the underlying nature of traffic. Specifically, the sensor networks in most traffic datasets do not accurately represent the actual road network exploited by vehicles, failing to provide insights into the traffic patterns in urban activities. To overcome these limitations, we propose an improved traffic prediction method based on graph convolution deep learning algorithms. We leverage human activity frequency data from National Household Travel Survey to enhance the inference capability of a causal relationship between activity and traffic patterns. Despite making minimal modifications to the conventional graph convolutional recurrent networks and graph convolutional transformer architectures, our approach achieves state-of-the-art performance without introducing excessive computational overhead.

Yun Kwan Kim, Minji Lee, Kunwook Jo et al.

Clinical decision support systems (CDSSs) have been widely utilized to support the decisions made by cardiologists when detecting and classifying arrhythmia from electrocardiograms (ECGs). However, forming a CDSS for the arrhythmia classification task is challenging due to the varying lengths of arrhythmias. Although the onset time of arrhythmia varies, previously developed methods have not considered such conditions. Thus, we propose a framework that consists of (i) local temporal information extraction, (ii) global pattern extraction, and (iii) local-global information fusion with attention to perform arrhythmia detection and classification with a constrained input length. The 10-class and 4-class performances of our approach were assessed by detecting the onset and offset of arrhythmia as an episode and the duration of arrhythmia based on the MIT-BIH arrhythmia database (MITDB) and MIT-BIH atrial fibrillation database (AFDB), respectively. The results were statistically superior to those achieved by the comparison models. To check the generalization ability of the proposed method, an AFDB-trained model was tested on the MITDB, and superior performance was attained compared with that of a state-of-the-art model. The proposed method can capture local-global information and dynamics without incurring information losses. Therefore, arrhythmias can be recognized more accurately, and their occurrence times can be calculated; thus, the clinical field can create more accurate treatment plans by using the proposed method.

Yeqing Lin, Minji Lee, Zhao Zhang et al.

Protein diffusion models have emerged as a promising approach for protein design. One such pioneering model is Genie, a method that asymmetrically represents protein structures during the forward and backward processes, using simple Gaussian noising for the former and expressive SE(3)-equivariant attention for the latter. In this work we introduce Genie 2, extending Genie to capture a larger and more diverse protein structure space through architectural innovations and massive data augmentation. Genie 2 adds motif scaffolding capabilities via a novel multi-motif framework that designs co-occurring motifs with unspecified inter-motif positions and orientations. This makes possible complex protein designs that engage multiple interaction partners and perform multiple functions. On both unconditional and conditional generation, Genie 2 achieves state-of-the-art performance, outperforming all known methods on key design metrics including designability, diversity, and novelty. Genie 2 also solves more motif scaffolding problems than other methods and does so with more unique and varied solutions. Taken together, these advances set a new standard for structure-based protein design. Genie 2 inference and training code, as well as model weights, are freely available at: https://github.com/aqlaboratory/genie2.

Luiz Felipe Vecchietti, Minji Lee, Begench Hangeldiyev et al.

Recent advancements in machine learning (ML) are transforming the field of structural biology. For example, AlphaFold, a groundbreaking neural network for protein structure prediction, has been widely adopted by researchers. The availability of easy-to-use interfaces and interpretable outcomes from the neural network architecture, such as the confidence scores used to color the predicted structures, have made AlphaFold accessible even to non-ML experts. In this paper, we present various methods for representing protein 3D structures from low- to high-resolution, and show how interpretable ML methods can support tasks such as predicting protein structures, protein function, and protein-protein interactions. This survey also emphasizes the significance of interpreting and visualizing ML-based inference for structure-based protein representations that enhance interpretability and knowledge discovery. Developing such interpretable approaches promises to further accelerate fields including drug development and protein design.

Minji Lee, Colin Kalicki, Minkyu Jeon et al.

Models from the AlphaFold (AF) family reliably predict one dominant conformation for most well-ordered proteins but struggle to capture biologically relevant alternate states. Several efforts have focused on eliciting greater conformational variability through ad hoc inference-time perturbations of AF models or their inputs. Despite their progress, these approaches remain inefficient and fail to consistently recover major conformational modes. Here, we investigate both the optimal location and manner-of-operation for perturbing latent representations in the AF3 architecture. We distill our findings in ConforNets: channel-wise affine transforms of the pre-Pairformer pair latents. Unlike previous methods, ConforNets globally modulate AF3 representations, making them reusable across proteins. On unsupervised generation of alternate states, ConforNets achieve state-of-the-art success rates on all existing multi-state benchmarks. On the novel supervised task of conformational transfer, ConforNets trained on one source protein can induce a conserved conformational change across a protein family. Collectively, these results introduce a mechanism for conformational control in AF3-based models.

Heejin Choi, Minji Lee, Chang Hyeong Lee et al.

In this work, we propose a new method for ordering nets during the process of layer assignment in global routing problems. The global routing problems that we focus on in this work are based on routing problems that occur in the design of substrates in multilayered semiconductor packages. The proposed new method is based on machine learning techniques and we show that the proposed method supersedes conventional net ordering techniques based on heuristic score functions. We perform global routing experiments in multilayered semiconductor package environments in order to illustrate that the routing order based on our new proposed technique outperforms previous methods based on heuristics. Our approach of using machine learning for global routing targets specifically the net ordering step which we show in this work can be significantly improved by deep learning.

Young-Min Go, Seong-Hyun Yu, Hyeong-Yeong Park et al.

Brain-computer interface (BCI) technology facilitates communication between the human brain and computers, primarily utilizing electroencephalography (EEG) signals to discern human intentions. Although EEG-based BCI systems have been developed for paralysis individuals, ongoing studies explore systems for speech imagery and motor imagery (MI). This study introduces FingerNet, a specialized network for fine MI classification, departing from conventional gross MI studies. The proposed FingerNet could extract spatial and temporal features from EEG signals, improving classification accuracy within the same hand. The experimental results demonstrated that performance showed significantly higher accuracy in classifying five finger-tapping tasks, encompassing thumb, index, middle, ring, and little finger movements. FingerNet demonstrated dominant performance compared to the conventional baseline models, EEGNet and DeepConvNet. The average accuracy for FingerNet was 0.3049, whereas EEGNet and DeepConvNet exhibited lower accuracies of 0.2196 and 0.2533, respectively. Statistical validation also demonstrates the predominance of FingerNet over baseline networks. For biased predictions, particularly for thumb and index classes, we led to the implementation of weighted cross-entropy and also adapted the weighted cross-entropy, a method conventionally employed to mitigate class imbalance. The proposed FingerNet involves optimizing network structure, improving performance, and exploring applications beyond fine MI. Moreover, the weighted Cross Entropy approach employed to address such biased predictions appears to have broader applicability and relevance across various domains involving multi-class classification tasks. We believe that effective execution of motor imagery can be achieved not only for fine MI, but also for local muscle MI

Sangwoo Youn, Minji Lee, Nokap Tony Park et al.

Video outpainting presents a unique challenge of extending the borders while maintaining consistency with the given content. In this paper, we suggest the use of video inpainting models that excel in object flow learning and reconstruction in outpainting rather than solely generating the background as in existing methods. However, directly applying or fine-tuning inpainting models to outpainting has shown to be ineffective, often leading to blurry results. Our extensive experiments on discriminator designs reveal that a critical component missing in the outpainting fine-tuning process is a discriminator capable of effectively assessing the perceptual quality of the extended areas. To tackle this limitation, we differentiate the objectives of adversarial training into global and local goals and introduce a hierarchical discriminator that meets both objectives. Additionally, we develop a specialized outpainting loss function that leverages both local and global features of the discriminator. Fine-tuning on this adversarial loss function enhances the generator's ability to produce both visually appealing and globally coherent outpainted scenes. Our proposed method outperforms state-of-the-art methods both quantitatively and qualitatively. Supplementary materials including the demo video and the code are available in SigPort.

Minji Lee

Multi-image super-resolution, which aims to fuse and restore a high-resolution image from multiple images at the same location, is crucial for utilizing satellite images. The satellite images are often occluded by atmospheric disturbances such as clouds, and the position of the disturbances varies by the images. Many radiometric and geometric approaches are proposed to detect atmospheric disturbances. Still, the utilization of detection results, i.e., quality maps in deep learning was limited to pre-processing or computation of loss. In this paper, we present a quality map-associated attention network (QA-Net), an architecture that fully incorporates QMs into a deep learning scheme for the first time. Our proposed attention modules process QMs alongside the low-resolution images and utilize the QM features to distinguish the disturbances and attend to image features. As a result, QA-Net has achieved state-of-the-art results in the PROBA-V dataset.

Jaemin Yoon, Minji Lee, Dongwon Son et al.

We propose a novel fast and accurate simulation framework for contact-intensive tight-tolerance robotic assembly tasks. The key components of our framework are as follows: 1) data-driven contact point clustering with a certain variable-input network, which is explicitly trained for simulation accuracy (with real experimental data) and able to accommodate complex/non-convex object shapes; 2) contact force solving, which precisely/robustly enforces physics of contact (i.e., no penetration, Coulomb friction, maximum energy dissipation) with contact mechanics of contact nodes augmented with that of their object; 3) contact detection with a neural network, which is parallelized for each contact point, thus, can be computed very quickly even for complex shape objects with no exhaust pair-wise test; and 4) time integration with PMI (passive mid-point integration), whose discrete-time passivity improves overall simulation accuracy, stability, and speed. We then implement our proposed framework for two widely-encountered/benchmarked contact-intensive tight-tolerance tasks, namely, peg-in-hole assembly and bolt-nut assembly, and validate its speed and accuracy against real experimental data. It is worthwhile to mention that our proposed simulation framework is applicable to other general contact-intensive tight-tolerance robotic assembly tasks as well. We also compare its performance with other physics engines and manifest its robustness via haptic rendering of virtual bolting task.

Jeongmin Lee, Minji Lee, Dongjun Lee

We propose a novel multibody dynamics simulation framework that can efficiently deal with large-dimensionality and complementarity multi-contact conditions. Typical contact simulation approaches perform contact impulse-level fixed-point iteration (IL-FPI), which has high time-complexity from large-size matrix inversion and multiplication, as well as susceptibility to ill-conditioned contact situations. To circumvent this, we propose a novel framework based on velocity-level fixed-point iteration (VL-FPI), which, by utilizing a certain surrogate dynamics and contact nodalization (with virtual nodes), can achieve not only inter-contact decoupling but also their inter-axes decoupling (i.e., contact diagonalization). This then enables us to one-shot/parallel-solve the contact problem during each VL-FPI iteration-loop, while the surrogate dynamics structure allows us to circumvent large-size/dense matrix inversion/multiplication, thereby, significantly speeding up the simulation time with improved convergence property. We theoretically show that the solution of our framework is consistent with that of the original problem and, further, elucidate mathematical conditions for the convergence of our proposed solver. Performance and properties of our proposed simulation framework are also demonstrated and experimentally-validated for various large-dimensional/multi-contact scenarios including deformable objects.

Young-Eun Lee, Gi-Hwan Shin, Minji Lee et al.

We present a mobile dataset obtained from electroencephalography (EEG) of the scalp and around the ear as well as from locomotion sensors by 24 participants moving at four different speeds while performing two brain-computer interface (BCI) tasks. The data were collected from 32-channel scalp-EEG, 14-channel ear-EEG, 4-channel electrooculography, and 9-channel inertial measurement units placed at the forehead, left ankle, and right ankle. The recording conditions were as follows: standing, slow walking, fast walking, and slight running at speeds of 0, 0.8, 1.6, and 2.0m/s, respectively. For each speed, two different BCI paradigms, event-related potential and steady-state visual evoked potential, were recorded. To evaluate the signal quality, scalp- and ear-EEG data were qualitatively and quantitatively validated during each speed. We believe that the dataset will facilitate BCIs in diverse mobile environments to analyze brain activities and evaluate the performance quantitatively for expanding the use of practical BCIs.

Young-Seok Kweon, Gi-Hwan Shin, Heon-Gyu Kwak et al.

A micro-sleep is a short sleep that lasts from 1 to 30 secs. Its detection during driving is crucial to prevent accidents that could claim a lot of people's lives. Electroencephalogram (EEG) is suitable to detect micro-sleep because EEG was associated with consciousness and sleep. Deep learning showed great performance in recognizing brain states, but sufficient data should be needed. However, collecting micro-sleep data during driving is inefficient and has a high risk of obtaining poor data quality due to noisy driving situations. Night-sleep data at home is easier to collect than micro-sleep data during driving. Therefore, we proposed a deep learning approach using night-sleep EEG to improve the performance of micro-sleep detection. We pre-trained the U-Net to classify the 5-class sleep stages using night-sleep EEG and used the sleep stages estimated by the U-Net to detect micro-sleep during driving. This improved micro-sleep detection performance by about 30\% compared to the traditional approach. Our approach was based on the hypothesis that micro-sleep corresponds to the early stage of non-rapid eye movement (NREM) sleep. We analyzed EEG distribution during night-sleep and micro-sleep and found that micro-sleep has a similar distribution to NREM sleep. Our results provide the possibility of similarity between micro-sleep and the early stage of NREM sleep and help prevent micro-sleep during driving.

Seo-Hyun Lee, Minji Lee, Seong-Whan Lee

Recent advances in brain-computer interface technology have shown the potential of imagined speech and visual imagery as a robust paradigm for intuitive brain-computer interface communication. However, the internal dynamics of the two paradigms along with their intrinsic features haven't been revealed. In this paper, we investigated the functional connectivity of the two paradigms, considering various frequency ranges. The dataset of sixteen subjects performing thirteen-class imagined speech and visual imagery were used for the analysis. The phase-locking value of imagined speech and visual imagery was analyzed in seven cortical regions with four frequency ranges. We compared the functional connectivity of imagined speech and visual imagery with the resting state to investigate the brain alterations during the imagery. The phase-locking value in the whole brain region exhibited a significant decrease during both imagined speech and visual imagery. Broca and Wernicke's area along with the auditory cortex mainly exhibited a significant decrease in the imagined speech, and the prefrontal cortex and the auditory cortex have shown a significant decrease in the visual imagery paradigm. Further investigation on the brain connectivity along with the decoding performance of the two paradigms may play a crucial role as a performance predictor.

Gi-Hwan Shin, Young-Seok Kweon, Minji Lee

Memory is an essential element in people's daily life based on experience. So far, many studies have analyzed electroencephalogram (EEG) signals at encoding to predict later remembered items, but few studies have predicted long-term memory only with EEG signals of successful short-term memory. Therefore, we aim to predict long-term memory using deep neural networks. In specific, the spectral power of the EEG signals of remembered items in short-term memory was calculated and inputted to the multilayer perceptron (MLP) and convolutional neural network (CNN) classifiers to predict long-term memory. Seventeen participants performed visuo-spatial memory task consisting of picture and location memory in the order of encoding, immediate retrieval (short-term memory), and delayed retrieval (long-term memory). We applied leave-one-subject-out cross-validation to evaluate the predictive models. As a result, the picture memory showed the highest kappa-value of 0.19 on CNN, and location memory showed the highest kappa-value of 0.32 in MLP. These results showed that long-term memory can be predicted with measured EEG signals during short-term memory, which improves learning efficiency and helps people with memory and cognitive impairments.

Young-Eun Lee, Minji Lee, Seong-Whan Lee

Practical brain-machine interfaces have been widely studied to accurately detect human intentions using brain signals in the real world. However, the electroencephalography (EEG) signals are distorted owing to the artifacts such as walking and head movement, so brain signals may be large in amplitude rather than desired EEG signals. Due to these artifacts, detecting accurately human intention in the mobile environment is challenging. In this paper, we proposed the reconstruction framework based on generative adversarial networks using the event-related potentials (ERP) during walking. We used a pre-trained convolutional encoder to represent latent variables and reconstructed ERP through the generative model which shape similar to the opposite of encoder. Finally, the ERP was classified using the discriminative model to demonstrate the validity of our proposed framework. As a result, the reconstructed signals had important components such as N200 and P300 similar to ERP during standing. The accuracy of reconstructed EEG was similar to raw noisy EEG signals during walking. The signal-to-noise ratio of reconstructed EEG was significantly increased as 1.3. The loss of the generative model was 0.6301, which is comparatively low, which means training generative model had high performance. The reconstructed ERP consequentially showed an improvement in classification performance during walking through the effects of noise reduction. The proposed framework could help recognize human intention based on the brain-machine interface even in the mobile environment.

Gi-Hwan Shin, Minji Lee, Seong-Whan Lee

The assessment of consciousness and unconsciousness is a challenging issue in modern neuroscience. Consciousness is closely related to memory consolidation in that memory is a critical component of conscious experience. So far, many studies have been reported on memory consolidation during consciousness, but there is little research on memory consolidation during unconsciousness. Therefore, we aim to assess the unconsciousness in terms of memory consolidation using electroencephalogram signals. In particular, we used unconscious state during a nap; because sleep is the only state in which consciousness disappears under normal physiological conditions. Seven participants performed two memory tasks (word-pairs and visuo-spatial) before and after the nap to assess the memory consolidation during unconsciousness. As a result, spindle power in central, parietal, occipital regions during unconsciousness was positively correlated with the performance of location memory. With the memory performance, there was also a negative correlation between delta connectivity and word-pairs memory, alpha connectivity and location memory, and spindle connectivity and word-pairs memory. We additionally observed the significant relationship between unconsciousness and brain changes during memory recall before and after the nap. These findings could help present new insights into the assessment of unconsciousness by exploring the relationship with memory consolidation.

Gi-Hwan Shin, Minji Lee, Hyeong-Jin Kim et al.

Event-related potential (ERP) speller can be utilized in device control and communication for locked-in or severely injured patients. However, problems such as inter-subject performance instability and ERP-illiteracy are still unresolved. Therefore, it is necessary to predict classification performance before performing an ERP speller in order to use it efficiently. In this study, we investigated the correlations with ERP speller performance using a resting-state before an ERP speller. In specific, we used spectral power and functional connectivity according to four brain regions and five frequency bands. As a result, the delta power in the frontal region and functional connectivity in the delta, alpha, gamma bands are significantly correlated with the ERP speller performance. Also, we predicted the ERP speller performance using EEG features in the resting-state. These findings may contribute to investigating the ERP-illiteracy and considering the appropriate alternatives for each user.

Jae-Geun Yoon, Minji Lee

Motor imagery based brain-computer interfaces (MI-BCIs) allow the control of devices and communication by imagining different muscle movements. However, most studies have reported a problem of "BCI-illiteracy" that does not have enough performance to use MI-BCI. Therefore, understanding subjects with poor performance and finding the cause of performance variation is still an important challenge. In this study, we proposed predictors of MI performance using effective connectivity in resting-state EEG. As a result, the high and low MI performance groups had a significant difference as 23% MI performance difference. We also found that connection from right lateral parietal to left lateral parietal in resting-state EEG was correlated significantly with MI performance (r = -0.37). These findings could help to understand BCI-illiteracy and to consider alternatives that are appropriate for the subject.

Jenifer Kalafatovich, Minji Lee

Declarative memory has been studied for its relationship with remembering daily life experiences. Previous studies reported changes in power spectra during encoding phase related to behavioral performance, however decoding phase still needs to be explored. This study investigates neural oscillations changes related to memory process. Participants were asked to perform a memory task for encoding and decoding phase while EEG signals were recorded. Results showed that for encoding phase, there was a significant decrease of power in low beta, high beta bands over fronto-central area and a decrease in low beta, high beta and gamma bands over left temporal area related to successful subsequent memory effects. For decoding phase, only significant decreases of alpha power were observed over fronto-central area. This finding showed relevance of beta and alpha band for encoding and decoding phase of a memory task respectively.

Seo-Hyun Lee, Minji Lee, Seong-Whan Lee

Visual imagery is an intuitive brain-computer interface paradigm, referring to the emergence of the visual scene. Despite its convenience, analysis of its intrinsic characteristics is limited. In this study, we demonstrate the effect of time interval and channel selection that affects the decoding performance of the multi-class visual imagery. We divided the epoch into time intervals of 0-1 s and 1-2 s and performed six-class classification in three different brain regions: whole brain, visual cortex, and prefrontal cortex. In the time interval, 0-1 s group showed 24.2 % of average classification accuracy, which was significantly higher than the 1-2 s group in the prefrontal cortex. In the three different regions, the classification accuracy of the prefrontal cortex showed significantly higher performance than the visual cortex in 0-1 s interval group, implying the cognitive arousal during the visual imagery. This finding would provide crucial information in improving the decoding performance.

Hyeong-Jin Kim, Min-Ho Lee, Minji Lee

Recently, smart home systems based on brain-computer interface (BCI) has attracted a wide range of interests in both industry and academia. However, the current BCI system has several shortcomings as it produces a comparatively lower accuracy for real-time implementations as well as the intuitive paradigm for the users cannot be well established here. Therefore, in this study, we proposed a highly intuitive BCI paradigm that combines event-related potential (ERP) with the speech-imagery task for the individual target objects. The decoding accuracy of the proposed paradigm was 88.1% (plus or minus 5.90) which is a much significant higher performance than a conventional ERP system. Furthermore, the amplitude of N700 components was significantly enhanced over frontal regions which are priory evoked by the speech-imagery task. Our results could be utilized to develop a smart home system so that it could be more user-friendly and convenient by means of delivering user's intentions both, intuitively and accurately.

Young-Eun Lee, Minji Lee

Recently, practical brain-computer interface is actively carried out, especially, in an ambulatory environment. However, the electroencephalography signals are distorted by movement artifacts and electromyography signals in ambulatory condition, which make hard to recognize human intention. In addition, as hardware issues are also challenging, ear-EEG has been developed for practical brain-computer interface and is widely used. However, ear-EEG still contains contaminated signals. In this paper, we proposed robust two-stream deep neural networks in walking conditions and analyzed the visual response EEG signals in the scalp and ear in terms of statistical analysis and brain-computer interface performance. We validated the signals with the visual response paradigm, steady-state visual evoked potential. The brain-computer interface performance deteriorated as 3~14% when walking fast at 1.6 m/s. When applying the proposed method, the accuracies increase 15% in cap-EEG and 7% in ear-EEG. The proposed method shows robust to the ambulatory condition in session dependent and session-to-session experiments.