arXiv subjects

Jinwoo Park

Publications and source records attributed to Jinwoo Park.

30 records · Page 2Linked to original sources

arXiv2024

Unveiling Key Aspects of Fine-Tuning in Sentence Embeddings: A Representation Rank Analysis

The latest advancements in unsupervised learning of sentence embeddings predominantly involve employing contrastive learning-based (CL-based) fine-tuning over pre-trained language models. In this study, we analyze the latest sentence embedding methods by adopting representation rank as the primary tool of analysis. We first define Phase 1 and Phase 2 of fine-tuning based on when representation rank peaks. Utilizing these phases, we conduct a thorough analysis and obtain essential findings across key aspects, including alignment and uniformity, linguistic abilities, and correlation between performance and rank. For instance, we find that the dynamics of the key aspects can undergo significant changes as fine-tuning transitions from Phase 1 to Phase 2. Based on these findings, we experiment with a rank reduction (RR) strategy that facilitates rapid and stable fine-tuning of the latest CL-based methods. Through empirical investigations, we showcase the efficacy of RR in enhancing the performance and stability of five state-of-the-art sentence embedding methods.

cs.CL↗

arXiv2021

Deep reinforcement learning for guidewire navigation in coronary artery phantom

In percutaneous intervention for treatment of coronary plaques, guidewire navigation is a primary procedure for stent delivery. Steering a flexible guidewire within coronary arteries requires considerable training, and the non-linearity between the control operation and the movement of the guidewire makes precise manipulation difficult. Here, we introduce a deep reinforcement learning(RL) framework for autonomous guidewire navigation in a robot-assisted coronary intervention. Using Rainbow, a segment-wise learning approach is applied to determine how best to accelerate training using human demonstrations with deep Q-learning from demonstrations (DQfD), transfer learning, and weight initialization. `State' for RL is customized as a focus window near the guidewire tip, and subgoals are placed to mitigate a sparse reward problem. The RL agent improves performance, eventually enabling the guidewire to reach all valid targets in `stable' phase. Our framework opens anew direction in the automation of robot-assisted intervention, providing guidance on RL in physical spaces involving mechanical fatigue.

cs.RO↗

arXiv2020

Clustering Approaches for Global Minimum Variance Portfolio

The only input to attain the portfolio weights of global minimum variance portfolio (GMVP) is the covariance matrix of returns of assets being considered for investment. Since the population covariance matrix is not known, investors use historical data to estimate it. Even though sample covariance matrix is an unbiased estimator of the population covariance matrix, it includes a great amount of estimation error especially when the number of observed data is not much bigger than number of assets. As it is difficult to estimate the covariance matrix with high dimensionality all at once, clustering stocks is proposed to come up with covariance matrix in two steps: firstly, within a cluster and secondly, between clusters. It decreases the estimation error by reducing the number of features in the data matrix. The motivation of this dissertation is that the estimation error can still remain high even after clustering, if a large amount of stocks is clustered together in a single group. This research proposes to utilize a bounded clustering method in order to limit the maximum cluster size. The result of experiments shows that not only the gap between in-sample volatility and out-of-sample volatility decreases, but also the out-of-sample volatility gets reduced. It implies that we need a bounded clustering algorithm so that maximum clustering size can be precisely controlled to find the best portfolio performance.

q-fin.PM↗

arXiv2020

Edge-functionalization of armchair graphene nanoribbons with pentagonal-hexagonal edge structures

Using density functional theory calculations, we have studied the edge-functionalization of armchair graphene nanoribbons (AGNRs) with pentagonal-hexagonal edge structures. While the AGNRs with pentagonal-hexagonal edge structures (labeled (5,6)-AGNRs) are metallic, the edge-functionalized (5,6)-AGNRs with substitutional atoms opens a band gap. We find that the band structures of edge-functionalized (5,6)-N-AGNRs by substitution resemble those of defect-free (N-1)-AGNR at the Γ point, whereas those at the X point show the original ones of the defect-free N-AGNR. The overall electronic structures of edge-functionalized (5,6)-AGNRs depend on the number of electrons, supplied by substitutional atoms, at the edges of functionalized (5,6)-AGNRs.

cond-mat.mes-hall↗

arXiv2016

High-throughput screening of metal-porphyrin-like graphenes for selective capture of carbon dioxide

Nano-materials, such as metal-organic frameworks, have been considered to capture CO$_2$. However, their application has been limited largely because they exhibit poor selectivity for flue gases and low capture capacity under low pressures. We perform a high-throughput screening for selective CO$_2$ capture from flue gases by using first principles thermodynamics. We find that elements with empty d orbitals selectively attract CO$_2$ from gaseous mixtures under low CO$_2$ pressures at 300 K and release it at ~450 K. CO$_2$ binding to elements involves hybridization of the metal d orbitals with the CO$_2$ $π$ orbitals and CO$_2$-transition metal complexes were observed in experiments. This result allows us to perform high-throughput screening to discover novel promising CO$_2$ capture materials with empty d orbitals and predict their capture performance under various conditions. Moreover, these findings provide physical insights into selective CO$_2$ capture and open a new path to explore CO$_2$ capture materials.

cond-mat.mtrl-sci↗

arXiv2015

Efficient noiseless linear amplification for light fields with larger amplitudes

We suggest and investigate a scheme for non-deterministic noiseless linear amplification of coherent states using successive photon addition, $(\hat a^{\dagger})^2$, where $\hat a^\dagger$ is the photon creation operator. We compare it with a previous proposal using the photon addition-then-subtraction, $\hat a \hat a^\dagger$, where $\hat a$ is the photon annihilation operator, that works as an appropriate amplifier only for weak light fields. We show that when the amplitude of a coherent state is $|α| \gtrsim 0.91$, the $(\hat a^{\dagger})^2$ operation serves as a more efficient amplifier compared to the $\hat a \hat a^\dagger$ operation in terms of equivalent input noise. Using $\hat a \hat a^\dagger$ and $(\hat a^{\dagger})^2$ as basic building blocks, we compare combinatorial amplifications of coherent states using $(\hat a \hat a^\dagger)^2$, $\hat a^{\dagger 4}$, $\hat a \hat a^\dagger\hat a^{\dagger 2}$, and $\hat a^{\dagger 2}\hat a \hat a^\dagger$, and show that $(\hat a \hat a^\dagger)^2$, $\hat a^{\dagger 2}\hat a \hat a^\dagger$, and $\hat a^{\dagger 4}$ exhibit strongest noiseless properties for $|α| \lesssim 0.51$, $0.51 \lesssim |α| \lesssim 1.05 $, and $|α|\gtrsim 1.05 $, respectively. We further show that the $(\hat a^{\dagger})^2$ operation can be used for amplifying superpositions of the coherent states. In contrast to previous studies, our work provides efficient schemes to implement a noiseless amplifier for light fields with medium and large amplitudes.

quant-ph↗

arXiv2015

Van der Waals density-functional theory study for bulk solids with BCC, FCC, and diamond structures

Proper inclusion of van der Waals (vdW) interactions in theoretical simulations based on standard density functional theory (DFT) is crucial to describe the physics and chemistry of systems such as organic and layered materials. Many encouraging approaches have been proposed to combine vdW interactions with standard approximate DFT calculations. Despite many vdW studies, there is no consensus on the reliability of vdW methods. To help further development of vdW methods, we have assessed various vdW functionals through the calculation of structural prop- erties at equilibrium, such as lattice constants, bulk moduli, and cohesive energies, for bulk solids, including alkali, alkali-earth, and transition metals, with BCC, FCC, and diamond structures as the ground state structure. These results provide important information for the vdW-related materials research, which is essential for designing and optimizing materials systems for desired physical and chemical properties.

cond-mat.mtrl-sci↗

arXiv2014

Transfer of different types of optical qubits over a lossy environment

We compare three different types of optical qubits for information transfer via quantum teleportation and direction transmission under photon losses. The three types of qubits are (1) qubits using the vacuum and the single-photon (VSP) states, (2) single-photon qubits using polarization degrees of freedom, i.e., polarized single-photon (PSP) qubits, and (3) coherent-state qubits that use two coherent states with opposite phases as the qubit basis. Our analysis shows that the teleportation scheme outperforms the direct transmission for most of cases as far as fidelities are concerned. Overall, VSP qubits are found to be the most efficient for both the direct transmission and teleportation under photon loss effects. The coherent-state qubits are more robust than PSP qubits either when their amplitudes are small as $|α| \lesssim 1.22$ or when photon loss effects are strong. Our results would provide useful and timely information for the development of practical optical quantum information processing particularly in the context of hybrid architectures.

quant-ph↗

arXiv2012

Bell inequality test with entanglement between an atom and a coherent state in a cavity

We study Bell inequality tests with entanglement between a coherent-state field in a cavity and a two-level atom. In order to detect the cavity field for such a test, photon on/off measurements and photon number parity measurements, respectively, are investigated. When photon on/off measurements are used, at least 50% of detec- tion efficiency is required to demonstrate violation of the Bell inequality. Photon number parity measurements for the cavity field can be effectively performed using ancillary atoms and an atomic detector, which leads to large degrees of Bell violations up to Cirel'son's bound. We also analyze decoherence effects in both field and atomic modes and discuss conditions required to perform a Bell inequality test free from the locality loophole.

quant-ph↗

arXiv2011

First-principles study of substitutional carbon pair and Stone-Wales defect complexes in boron nitride nanotubes

Using density functional theory, we study physical properties of boron nitride nanotubes (BNNTs) with the substitutional carbon pair defect. We also consider the Stone-Wales (SW) rearrangement of the C-C pair defect in the BNNT. The formation energy of an SW defect of the carbon dimer is approximately 3.1 eV lower than that of the SW-transformed B-N pair in the undoped BNNT. The activation energies show that the SW defect in the C-doped BNNT may be experimentally observed with a higher probability than in the undoped BNNT. Finally, we discuss the localized states originating from the carbon pair impurities.

cond-mat.mes-hall↗

arXiv2011

A Real-time Localization System Using RFID for Visually Impaired

Gadgets helping the disabled, especially blind that are in least accessibility of information, use acoustic methods that can cause stress to ear and infringe user's privacy. Even if some project uses embedded Radio Frequency Identification (RFID) into the sidewalk for blind's free walking, the tag memory design is not specified for buildings and road conditions. This paper suggested allocation scheme of RFID tag referring to EPCglobal SGLN, tactile method for conveying information, and use of lithium battery as power source with solar cells as an alternative. Results have shown independent mobility, accidents prevention, stress relief and satisfied factors in terms of cost and human usability.

cs.MA↗

arXiv2010

Quasi-dark Mode in a Metamaterial for Analogous Electromagnetically-induced Transparency

We study a planar metamaterial supporting electromagnetically-induced transparency (EIT)-like effect by exploiting the coupling between bright and quasi-dark eigenmodes. The specific design of such a metamaterial consists of a cut-wire (CW) and a single-gap split-ring resonator (SRR). From the numerical and the analytical results we demonstrate that the response of SRR, which is weakly excited by external electric field, is mitigated to be a quasi-dark eigenmode in the presence of strongly radiative CW. This result suggests more relaxed conditions for the realization of devices utilizing the EIT-like effects in metamaterial, and thereby widens the possibilities for many different structural implementations.

physics.optics↗