arXiv subjects

Seunghyun Lee

Publications and source records attributed to Seunghyun Lee.

At least 73 records · Page 4Linked to original sources

arXiv2023

Clinical Decision Transformer: Intended Treatment Recommendation through Goal Prompting

With recent achievements in tasks requiring context awareness, foundation models have been adopted to treat large-scale data from electronic health record (EHR) systems. However, previous clinical recommender systems based on foundation models have a limited purpose of imitating clinicians' behavior and do not directly consider a problem of missing values. In this paper, we propose Clinical Decision Transformer (CDT), a recommender system that generates a sequence of medications to reach a desired range of clinical states given as goal prompts. For this, we conducted goal-conditioned sequencing, which generated a subsequence of treatment history with prepended future goal state, and trained the CDT to model sequential medications required to reach that goal state. For contextual embedding over intra-admission and inter-admissions, we adopted a GPT-based architecture with an admission-wise attention mask and column embedding. In an experiment, we extracted a diabetes dataset from an EHR system, which contained treatment histories of 4788 patients. We observed that the CDT achieved the intended treatment effect according to goal prompt ranges (e.g., NormalA1c, LowerA1c, and HigherA1c), contrary to the case with behavior cloning. To the best of our knowledge, this is the first study to explore clinical recommendations from the perspective of goal prompting. See https://clinical-decision-transformer.github.io for code and additional information.

cs.AI↗

arXiv2022

Evaluating the Knowledge Dependency of Questions

The automatic generation of Multiple Choice Questions (MCQ) has the potential to reduce the time educators spend on student assessment significantly. However, existing evaluation metrics for MCQ generation, such as BLEU, ROUGE, and METEOR, focus on the n-gram based similarity of the generated MCQ to the gold sample in the dataset and disregard their educational value. They fail to evaluate the MCQ's ability to assess the student's knowledge of the corresponding target fact. To tackle this issue, we propose a novel automatic evaluation metric, coined Knowledge Dependent Answerability (KDA), which measures the MCQ's answerability given knowledge of the target fact. Specifically, we first show how to measure KDA based on student responses from a human survey. Then, we propose two automatic evaluation metrics, KDA_disc and KDA_cont, that approximate KDA by leveraging pre-trained language models to imitate students' problem-solving behavior. Through our human studies, we show that KDA_disc and KDA_soft have strong correlations with both (1) KDA and (2) usability in an actual classroom setting, labeled by experts. Furthermore, when combined with n-gram based similarity metrics, KDA_disc and KDA_cont are shown to have a strong predictive power for various expert-labeled MCQ quality measures.

cs.CL↗

arXiv2022

Autonomous Delivery of Multiple Packages Using Single Drone in Urban Airspace

Current drone delivery solutions mainly focus on single package delivery using one drone. However, the recent developments in drone technology enable a drone to deliver multiple packages in a single trip. We use the nearest destination first strategy for the faster delivery of packages in a skyway network. This demonstration is a proof-of-concept prototype for the multi-package delivery in urban airspace following a skyway network. We deploy and test this multi-package drone delivery in an indoor testbed environment using a 3D model of Sydney CBD. Demo: https://youtu.be/YTwsIfUvWPc

cs.RO↗

arXiv2022

Ensemble Knowledge Guided Sub-network Search and Fine-tuning for Filter Pruning

Conventional NAS-based pruning algorithms aim to find the sub-network with the best validation performance. However, validation performance does not successfully represent test performance, i.e., potential performance. Also, although fine-tuning the pruned network to restore the performance drop is an inevitable process, few studies have handled this issue. This paper provides a novel Ensemble Knowledge Guidance (EKG) to solve both problems at once. First, we experimentally prove that the fluctuation of loss landscape can be an effective metric to evaluate the potential performance. In order to search a sub-network with the smoothest loss landscape at a low cost, we employ EKG as a search reward. EKG utilized for the following search iteration is composed of the ensemble knowledge of interim sub-networks, i.e., the by-products of the sub-network evaluation. Next, we reuse EKG to provide a gentle and informative guidance to the pruned network while fine-tuning the pruned network. Since EKG is implemented as a memory bank in both phases, it requires a negligible cost. For example, when pruning and training ResNet-50, just 315 GPU hours are required to remove around 45.04% of FLOPS without any performance degradation, which can operate even on a low-spec workstation. the implemented code is available at https://github.com/sseung0703/EKG.

cs.LG↗

arXiv2022

CFA: Coupled-hypersphere-based Feature Adaptation for Target-Oriented Anomaly Localization

For a long time, anomaly localization has been widely used in industries. Previous studies focused on approximating the distribution of normal features without adaptation to a target dataset. However, since anomaly localization should precisely discriminate normal and abnormal features, the absence of adaptation may make the normality of abnormal features overestimated. Thus, we propose Coupled-hypersphere-based Feature Adaptation (CFA) which accomplishes sophisticated anomaly localization using features adapted to the target dataset. CFA consists of (1) a learnable patch descriptor that learns and embeds target-oriented features and (2) scalable memory bank independent of the size of the target dataset. And, CFA adopts transfer learning to increase the normal feature density so that abnormal features can be clearly distinguished by applying patch descriptor and memory bank to a pre-trained CNN. The proposed method outperforms the previous methods quantitatively and qualitatively. For example, it provides an AUROC score of 99.5% in anomaly detection and 98.5% in anomaly localization of MVTec AD benchmark. In addition, this paper points out the negative effects of biased features of pre-trained CNNs and emphasizes the importance of the adaptation to the target dataset. The code is publicly available at https://github.com/sungwool/CFA_for_anomaly_localization.

cs.CV↗

arXiv2022

FlexBlock: A Flexible DNN Training Accelerator with Multi-Mode Block Floating Point Support

Training deep neural networks (DNNs) is a computationally expensive job, which can take weeks or months even with high performance GPUs. As a remedy for this challenge, community has started exploring the use of more efficient data representations in the training process, e.g., block floating point (BFP). However, prior work on BFP-based DNN accelerators rely on a specific BFP representation making them less versatile. This paper builds upon an algorithmic observation that we can accelerate the training by leveraging multiple BFP precisions without compromising the finally achieved accuracy. Backed up by this algorithmic opportunity, we develop a flexible DNN training accelerator, dubbed FlexBlock, which supports three different BFP precision modes, possibly different among activation, weight, and gradient tensors. While several prior works proposed such multi-precision support for DNN accelerators, not only do they focus only on the inference, but also their core utilization is suboptimal at a fixed precision and specific layer types when the training is considered. Instead, FlexBlock is designed in such a way that high core utilization is achievable for i) various layer types, and ii) three BFP precisions by mapping data in a hierarchical manner to its compute units. We evaluate the effectiveness of FlexBlock architecture using well-known DNNs on CIFAR, ImageNet and WMT14 datasets. As a result, training in FlexBlock significantly improves the training speed by 1.5~5.3x and the energy efficiency by 2.4~7.0x on average compared to other training accelerators and incurs marginal accuracy loss compared to full-precision training.

cs.LG↗

arXiv2021

Vision Transformer for Small-Size Datasets

Recently, the Vision Transformer (ViT), which applied the transformer structure to the image classification task, has outperformed convolutional neural networks. However, the high performance of the ViT results from pre-training using a large-size dataset such as JFT-300M, and its dependence on a large dataset is interpreted as due to low locality inductive bias. This paper proposes Shifted Patch Tokenization (SPT) and Locality Self-Attention (LSA), which effectively solve the lack of locality inductive bias and enable it to learn from scratch even on small-size datasets. Moreover, SPT and LSA are generic and effective add-on modules that are easily applicable to various ViTs. Experimental results show that when both SPT and LSA were applied to the ViTs, the performance improved by an average of 2.96% in Tiny-ImageNet, which is a representative small-size dataset. Especially, Swin Transformer achieved an overwhelming performance improvement of 4.08% thanks to the proposed SPT and LSA.

cs.CV↗

arXiv2021

Offline-to-Online Reinforcement Learning via Balanced Replay and Pessimistic Q-Ensemble

Recent advance in deep offline reinforcement learning (RL) has made it possible to train strong robotic agents from offline datasets. However, depending on the quality of the trained agents and the application being considered, it is often desirable to fine-tune such agents via further online interactions. In this paper, we observe that state-action distribution shift may lead to severe bootstrap error during fine-tuning, which destroys the good initial policy obtained via offline RL. To address this issue, we first propose a balanced replay scheme that prioritizes samples encountered online while also encouraging the use of near-on-policy samples from the offline dataset. Furthermore, we leverage multiple Q-functions trained pessimistically offline, thereby preventing overoptimism concerning unfamiliar actions at novel states during the initial training phase. We show that the proposed method improves sample-efficiency and final performance of the fine-tuned robotic agents on various locomotion and manipulation tasks. Our code is available at: https://github.com/shlee94/Off2OnRL.

cs.RO↗

arXiv2021

Contextual Gradient Scaling for Few-Shot Learning

Model-agnostic meta-learning (MAML) is a well-known optimization-based meta-learning algorithm that works well in various computer vision tasks, e.g., few-shot classification. MAML is to learn an initialization so that a model can adapt to a new task in a few steps. However, since the gradient norm of a classifier (head) is much bigger than those of backbone layers, the model focuses on learning the decision boundary of the classifier with similar representations. Furthermore, gradient norms of high-level layers are small than those of the other layers. So, the backbone of MAML usually learns task-generic features, which results in deteriorated adaptation performance in the inner-loop. To resolve or mitigate this problem, we propose contextual gradient scaling (CxGrad), which scales gradient norms of the backbone to facilitate learning task-specific knowledge in the inner-loop. Since the scaling factors are generated from task-conditioned parameters, gradient norms of the backbone can be scaled in a task-wise fashion. Experimental results show that CxGrad effectively encourages the backbone to learn task-specific knowledge in the inner-loop and improves the performance of MAML up to a significant margin in both same- and cross-domain few-shot classification.

cs.CV↗

arXiv2021

Conditions for bubbles to arise under heterogeneous beliefs

This paper studies the equilibrium price of a continuous time asset traded in a market with heterogeneous investors. We consider a positive mean reverting asset and two groups of investors who have different beliefs on the speed of mean reversion and the mean level. We provide an equivalent condition for bubbles to exist and show that price bubbles may not form even though there are heterogeneous beliefs. This condition is directly related to the drift term of the asset. In addition, we characterize the minimal equilibrium price as a unique $C^2$ solution of a differential equation and express it using confluent hypergeometric functions.

q-fin.MF↗

arXiv2021

Interpretable Embedding Procedure Knowledge Transfer via Stacked Principal Component Analysis and Graph Neural Network

Knowledge distillation (KD) is one of the most useful techniques for light-weight neural networks. Although neural networks have a clear purpose of embedding datasets into the low-dimensional space, the existing knowledge was quite far from this purpose and provided only limited information. We argue that good knowledge should be able to interpret the embedding procedure. This paper proposes a method of generating interpretable embedding procedure (IEP) knowledge based on principal component analysis, and distilling it based on a message passing neural network. Experimental results show that the student network trained by the proposed KD method improves 2.28% in the CIFAR100 dataset, which is higher performance than the state-of-the-art (SOTA) method. We also demonstrate that the embedding procedure knowledge is interpretable via visualization of the proposed KD process. The implemented code is available at https://github.com/sseung0703/IEPKT.

cs.CV↗

arXiv2020

LinkedIn's Audience Engagements API: A Privacy Preserving Data Analytics System at Scale

We present a privacy system that leverages differential privacy to protect LinkedIn members' data while also providing audience engagement insights to enable marketing analytics related applications. We detail the differentially private algorithms and other privacy safeguards used to provide results that can be used with existing real-time data analytics platforms, specifically with the open sourced Pinot system. Our privacy system provides user-level privacy guarantees. As part of our privacy system, we include a budget management service that enforces a strict differential privacy budget on the returned results to the analyst. This budget management service brings together the latest research in differential privacy into a product to maintain utility given a fixed differential privacy budget.

cs.CR↗

arXiv2020

Temperature-dependent rotationally inelastic collisions of OH- and He

We have studied the fundamental rotational relaxation and excitation collision of OH- J=0 <-> 1 with helium at different collision energies. Using state-selected photodetachment in a cryogenic ion trap, the collisional excitation of the first excited rotational state of OH- has been investigated and absolute inelastic collision rate coefficients have been extracted for collision temperatures between 20 and 35 K. The rates are compared with accurate quantum scattering calculations for three different potential energy surfaces. Good agreement is found within the experimental accuracy, but the experimental trend of increasing collision rates with temperature is only in part reflected in the calculations.

physics.atom-ph↗

arXiv2020

Context-aware Dynamics Model for Generalization in Model-Based Reinforcement Learning

Model-based reinforcement learning (RL) enjoys several benefits, such as data-efficiency and planning, by learning a model of the environment's dynamics. However, learning a global model that can generalize across different dynamics is a challenging task. To tackle this problem, we decompose the task of learning a global dynamics model into two stages: (a) learning a context latent vector that captures the local dynamics, then (b) predicting the next state conditioned on it. In order to encode dynamics-specific information into the context latent vector, we introduce a novel loss function that encourages the context latent vector to be useful for predicting both forward and backward dynamics. The proposed method achieves superior generalization ability across various simulated robotics and control tasks, compared to existing RL schemes.

cs.LG↗

arXiv2019

Graph-based Knowledge Distillation by Multi-head Attention Network

Knowledge distillation (KD) is a technique to derive optimal performance from a small student network (SN) by distilling knowledge of a large teacher network (TN) and transferring the distilled knowledge to the small SN. Since a role of convolutional neural network (CNN) in KD is to embed a dataset so as to perform a given task well, it is very important to acquire knowledge that considers intra-data relations. Conventional KD methods have concentrated on distilling knowledge in data units. To our knowledge, any KD methods for distilling information in dataset units have not yet been proposed. Therefore, this paper proposes a novel method that enables distillation of dataset-based knowledge from the TN using an attention network. The knowledge of the embedding procedure of the TN is distilled to graph by multi-head attention (MHA), and multi-task learning is performed to give relational inductive bias to the SN. The MHA can provide clear information about the source dataset, which can greatly improves the performance of the SN. Experimental results show that the proposed method is 7.05% higher than the SN alone for CIFAR100, which is 2.46% higher than the state-of-the-art.

cs.LG↗

arXiv2017

Retrodirective Multi-User Wireless Power Transfer with Massive MIMO

This letter studies a radio-frequency (RF) multi-user wireless power transfer (WPT) system, where an energy transmitter (ET) with a large number of antennas delivers energy wirelessly to multiple distributed energy receivers (ERs). We investigate a low-complexity WPT scheme based on the retrodirective beamforming technique, where all ERs send a common beacon signal simultaneously to the ET in the uplink and the ET simply conjugates and amplifies its received sum-signal and transmits to all ERs in the downlink for WPT. We show that such a low-complexity scheme achieves the massive multiple-input multiple-output (MIMO) energy beamforming gain. However, a "doubly near-far" issue exists due to the round-trip (uplink beacon and downlink WPT) signal propagation loss where the harvested power of a far ER from the ET can be significantly lower than that of a near ER if the same uplink beacon power is used. To tackle this problem, we propose a distributed uplink beacon power update algorithm, where each ER independently adjusts the beacon power based on its current harvested power in an iterative manner. It is shown that the proposed algorithm converges quickly to a unique fixed-point solution, which helps achieve the desired user fairness with best efforts.

cs.IT↗

arXiv2016

Evidence of electronic cloaking from chiral electron transport in bilayer graphene nanostructures

The coupling of charge carrier motion and pseudospin via chirality for massless Dirac fermions in monolayer graphene has generated dramatic consequences, such as the unusual quantum Hall effect and Klein tunneling. In bilayer graphene, charge carriers are massive Dirac fermions with a finite density of states at zero energy. Because of their non-relativistic nature, massive Dirac fermions can provide an even better test bed with which to clarify the importance of chirality in transport measurement than massless Dirac fermions in monolayer graphene. Here, we report an electronic cloaking effect as a manifestation of chirality by probing phase coherent transport in chemical-vapor-deposited bilayer graphene. Conductance oscillations with different periodicities were observed on extremely narrow bilayer graphene heterojunctions through electrostatic gating. Using a Fourier analysis technique, we identified the origin of each individual interference pattern. Importantly, the electron waves on the two sides of the potential barrier can be coupled through the evanescent waves inside the barrier, making the confined states underneath the barrier invisible to electrons. These findings provide direct evidence for the electronic cloaking effect and hold promise for the realization of pseudospintronics based on bilayer graphene.

cond-mat.mes-hall↗

arXiv2016

Distributed Wireless Power Transfer with Energy Feedback

Energy beamforming (EB) is a key technique for achieving efficient radio-frequency (RF) transmission enabled wireless energy transfer (WET). By optimally designing the waveforms from multiple energy transmitters (ETs) over the wireless channels, they can be constructively combined at the energy receiver (ER) to achieve an EB gain that scales with the number of ETs. However, the optimal design of EB waveforms requires accurate channel state information (CSI) at the ETs, which is challenging to obtain practically, especially in a distributed system with ETs at separate locations. In this paper, we study practical and efficient channel training methods to achieve optimal EB in a distributed WET system. We propose two protocols with and without centralized coordination, respectively, where distributed ETs either sequentially or in parallel adapt their transmit phases based on a low-complexity energy feedback from the ER. The energy feedback only depends on the received power level at the ER, where each feedback indicates one particular transmit phase that results in the maximum harvested power over a set of previously used phases. Simulation results show that the two proposed training protocols converge very fast in practical WET systems even with a large number of distributed ETs, while the protocol with sequential ET phase adaptation is also analytically shown to converge to the optimal EB design with perfect CSI by increasing the training time. Numerical results are also provided to evaluate the performance of the proposed distributed EB and training designs as compared to other benchmark schemes.

cs.IT↗