BZPEERPapers, people, evidence and research paths
Menu

Discover

SearchFind papers, people, topics, institutions and opportunities from one place.GraphSee how papers, authors, topics and evidence connect.

Research tools

HomeContinue from the papers, people and topics that matter now.CollectionsTurn saved papers into reading lists, evidence maps and shareable context.ResearchReconstruct the path that led you from search to paper to question.CollaborationMove from a paper, topic or expert request into a focused thread.PublishCreate full-text research articles with metadata, formulas and versions.

Network

ExpertsFind specialists whose work and reviews explain why they can help.CommunitiesJoin research groups organized around papers, topics and shared evidence.

Opportunities

ListingsBrowse roles, calls and collaborations connected to real research context.

Account

Log inReturn to your saved papers, graph views and active threads.Create accountStart saving papers, building a research trail and joining teams.
Log inCreate account

Source author record

Changwook Jeong

Changwook Jeong appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record
cond-mat.mes-hallMachine LearningApplicationscond-mat.mtrl-sciArtificial Intelligenceeess.SPphysics.comp-ph

Catalog footprint

What is connected

9works
7topics
4close collaborators

Actions

Connect this record

Log in to claim
Open graphBrowse works

Research graph

See the researcher in context

Open full explorer

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

9 published item(s)

preprint2022arXiv

Learning Student-Friendly Teacher Networks for Knowledge Distillation

We propose a novel knowledge distillation approach to facilitate the transfer of dark knowledge from a teacher to a student. Contrary to most of the existing methods that rely on effective training of student models given pretrained teachers, we aim to learn the teacher models that are friendly to students and, consequently, more appropriate for knowledge transfer. In other words, at the time of optimizing a teacher model, the proposed algorithm learns the student branches jointly to obtain student-friendly representations. Since the main goal of our approach lies in training teacher models and the subsequent knowledge distillation procedure is straightforward, most of the existing knowledge distillation methods can adopt this technique to improve the performance of diverse student models in terms of accuracy and convergence speed. The proposed algorithm demonstrates outstanding accuracy in several well-known knowledge distillation techniques with various combinations of teacher and student models even in the case that their architectures are heterogeneous and there is no prior knowledge about student models at the time of training teacher networks.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

PAC-Net: A Model Pruning Approach to Inductive Transfer Learning

Inductive transfer learning aims to learn from a small amount of training data for the target task by utilizing a pre-trained model from the source task. Most strategies that involve large-scale deep learning models adopt initialization with the pre-trained model and fine-tuning for the target task. However, when using over-parameterized models, we can often prune the model without sacrificing the accuracy of the source task. This motivates us to adopt model pruning for transfer learning with deep learning models. In this paper, we propose PAC-Net, a simple yet effective approach for transfer learning based on pruning. PAC-Net consists of three steps: Prune, Allocate, and Calibrate (PAC). The main idea behind these steps is to identify essential weights for the source task, fine-tune on the source task by updating the essential weights, and then calibrate on the target task by updating the remaining redundant weights. Under the various and extensive set of inductive transfer learning experiments, we show that our method achieves state-of-the-art performance by a large margin.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Restructuring TCAD System: Teaching Traditional TCAD New Tricks

Traditional TCAD simulation has succeeded in predicting and optimizing the device performance; however, it still faces a massive challenge - a high computational cost. There have been many attempts to replace TCAD with deep learning, but it has not yet been completely replaced. This paper presents a novel algorithm restructuring the traditional TCAD system. The proposed algorithm predicts three-dimensional (3-D) TCAD simulation in real-time while capturing a variance, enables deep learning and TCAD to complement each other, and fully resolves convergence errors.

0 citations0 reviewsNo code linkedOpen access
preprint2011arXiv

An efficient algorithm to calculate intrinsic thermoelectric parameters based on Landauer approach

The Landauer approach provides a conceptually simple way to calculate the intrinsic thermoelectric (TE) parameters of materials from the ballistic to the diffusive transport regime. This method relies on the calculation of the number of propagating modes and the scattering rate for each mode. The modes are calculated from the energy dispersion (E(k)) of the materials which require heavy computation and often supply energy relation on sparse momentum (k) grids. Here an efficient method to calculate the distribution of modes (DOM) from a given E(k) relationship is presented. The main features of this algorithm are, (i) its ability to work on sparse dispersion data, and (ii) creation of an energy grid for the DOM that is almost independent of the dispersion data therefore allowing for efficient and fast calculation of TE parameters. The inclusion of scattering effects is also straight forward. The effect of k-grid sparsity on the compute time for DOM and on the sensitivity of the calculated TE results are provided. The algorithm calculates the TE parameters within 5% accuracy when the K-grid sparsity is increased up to 60% for all the dimensions (3D, 2D and 1D). The time taken for the DOM calculation is strongly influenced by the transverse K density (K perpendicular to transport direction) but is almost independent of the transport K density (along the transport direction). The DOM and TE results from the algorithm are bench-marked with, (i) analytical calculations for parabolic bands, and (ii) realistic electronic and phonon results for $Bi_{2}Te_{3}$.

0 citations0 reviewsNo code linkedOpen access
preprint2011arXiv

On the Best Bandstructure for Thermoelectric Performance

The conventional understanding that a bandstructure that produces a Dirac delta function transport distribution (or transmission in the Landauer framework) maximizes the thermoelectric figure of merit, ZT, is revisited. Thermoelectric (TE) performance is evaluated using a simple tight binding (TB) model for electron dispersion and three different scattering models: 1) a constant scattering time, 2) a constant mean-free-path, and 3) a scattering rate proportional to the density-of-states. We found that a Dirac delta-function transmission never produces the maximum ZT. The best bandstructure for maximizing ZT depends on the scattering physics. These results demonstrate that the selection of bandstructure to maximize TE performance is more complex than previously thought and that a high density-of-states near the band edge does not necessarily improve TE performance.

0 citations0 reviewsNo code linkedOpen access
preprint2011arXiv

Relating electronic structure to thermoelectric device performance

Realistic thermoelectric modeling and simulation tools are needed to explain the experiments and for device design. In this paper, we present a simple computational technique to make use of rigorous band structure calculations in thermoelectric device design. Our, methodology, based on Landauer theory, provides a way to benchmark effective mass level models against rigorous band structures. It can also be used for determining the temperature-dependent Fermi-level from a given doping density and numerically generated density-of-states, which can then be used with the numerically generated distribution of channels to evaluate the temperature dependent electrical conductivity, thermopower, and electronic thermal conductivity. We illustrate the technique for silicon, for which well-calibrated band structure and transport data is available, but the technique is more generally suited to complex thermoelectric materials and provides an easy way to make use of first principles band structure calculations in device design.

0 citations0 reviewsNo code linkedOpen access
preprint2010arXiv

Landauer vs. Boltzmann and Full Dispersion vs. Debye Model Evaluation of Lattice Thermal Conductivity

Using a full dispersion description of phonons, the thermal conductivities of bulk Si and Bi2Te3 are evaluated using a Landauer approach and related to the conventional approach based on the Boltzmann transport equation. A procedure to extract a well-defined average phonon mean-free-path from the measured thermal conductivity and given phonon-dispersion is presented. The extracted mean-free-path has strong physical significance and differs greatly from simple estimates. The use of simplified dispersion models for phonons is discussed, and it is shown that two different Debye temperatures must be used to treat the specific heat and thermal conductivity (analogous to the two different effective masses needed to describe the electron density and conductivity). A simple technique to extract these two Debye temperatures is presented and the limitations of the method are discussed.

0 citations0 reviewsNo code linkedOpen access
preprint2009arXiv

On Landauer vs. Boltzmann and Full Band vs. Effective Mass Evaluation of Thermoelectric Transport Coefficients

The Landauer approach to diffusive transport is mathematically related to the solution of the Boltzmann transport equation, and expressions for the thermoelectric parameters in both formalisms are presented. Quantum mechanical and semiclassical techniques to obtain from a full description of the bandstructure, E(k), the number of conducting channels in the Landauer approach or the transport distribution in the Boltzmann solution are developed and compared. Thermoelectric transport coefficients are evaluated from an atomistic level, full band description of a crystal. Several example calculations for representative bulk materials are presented, and the full band results are related to the more common effective mass formalism. Finally, given a full E(k) for a crystal, a procedure to extract an accurate, effective mass level description is presented.

0 citations0 reviewsNo code linkedOpen access
preprint2009arXiv

On momentum conservation and thermionic emission cooling

The question of whether relaxing momentum conservation can increase the performance of thermionic cooling device is examined. Both homojunctions and heterojunctions are considered. It is shown that for many cases, a non-conserved lateral momentum model overestimates the current. For the case of heterojunctions with a much heavier effective mass in the barrier and with a low barrier height, however, non-conservation of lateral momentum may increase the current. These results may be simply understood from the general principle that the current is limited by the location, well or barrier, with the smallest number of conducting channels. These results also show that within thermionic emission framework, the possibilities of increasing thermionic cooling by relaxing momentum conservation are limited. More generally, however, when the connection to the source is weak or in the presence of scattering, the situation may be different. Issues that deserve further study are identified.

0 citations0 reviewsNo code linkedOpen access