Yeongjong Kim
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Yeongjong Kim has a total of 2 co-authors. Here are the top co-authors by number of shared publications:
- Dabeen Lee
Department of Industrial and Systems Engineering KAIST 34141 Daejeon Republic of Korea
1 shared publication - Sanghoon Baek
KAIST
1 shared publication
Latest Publications
- Open Access
Stochastic-Constrained Stochastic Optimization with Markovian Data
This paper considers stochastic-constrained stochastic optimization where the stochastic constraint is to satisfy that the expectation of a random function is below a certain threshold. In particular, we study the setting where data samples are drawn from a Markov chain and thus are not independent and identically distributed. We generalize the drift-plus-penalty framework, a primal-dual stochastic gradient method developed for the i.i.d. case, to the Markov chain sampling setting. We propose two variants of drift-plus-penalty; one is for the case when the mixing time of the underlying Markov chain is known while the other is for the case of unknown mixing time. In fact, our algorithms apply to a more general setting of constrained online convex optimization where the sequence of constraint functions follows a Markov chain. Both algorithms are adaptive in that the first works without knowledge of the time horizon while the second uses AdaGrad-style algorithm parameters, which is of independent interest. We demonstrate the effectiveness of our proposed methods through numerical experiments on classification with fairness constraints.
Open AccessStochastic-Constrained Stochastic Optimization with Markovian Data
This paper considers stochastic-constrained stochastic optimization where the stochastic constraint is to satisfy that the expectation of a random function is below a certain threshold. In particular, we study the setting where data samples are drawn from a Markov chain and thus are not independent and identically distributed. We generalize the drift-plus-penalty framework, a primal-dual stochastic gradient method developed for the i.i.d. case, to the Markov chain sampling setting. We propose two variants of drift-plus-penalty; one is for the case when the mixing time of the underlying Markov chain is known while the other is for the case of unknown mixing time. In fact, our algorithms apply to a more general setting of constrained online convex optimization where the sequence of constraint functions follows a Markov chain. Both algorithms are adaptive in that the first works without knowledge of the time horizon while the second uses AdaGrad-style algorithm parameters, which is of independent interest. We demonstrate the effectiveness of our proposed methods through numerical experiments on classification with fairness constraints.
- Open Access
Essential dimension of reductive groups via generically free representations
We provide a simple method to compute upper bounds on the essential dimension of split reductive groups with finite or connected center by means of their generically free representations. Combining our upper bound with previously known lower bound, the exact value of the essential dimension is calculated for some types of reductive groups. As an application, we determine the essential dimension of a semisimple group of classical type or $E_{6}$, and its strict reductive envelope under certain conditions on its center. This extends previous works on simple simply connected groups of type $B$ or $D$ by Brosnan-Reichstein-Vistoli and Chernousov-Merkurjev, strict reductive envelopes of groups of type $A$ by Cernele-Reichstein, and semisimple groups of type $B$ by the authors to any classical type and type $E_{6}$ in a uniform way.Open AccessEssential dimension of reductive groups via generically free representations
We provide a simple method to compute upper bounds on the essential dimension of split reductive groups with finite or connected center by means of their generically free representations. Combining our upper bound with previously known lower bound, the exact value of the essential dimension is calculated for some types of reductive groups. As an application, we determine the essential dimension of a semisimple group of classical type or $E_{6}$, and its strict reductive envelope under certain conditions on its center. This extends previous works on simple simply connected groups of type $B$ or $D$ by Brosnan-Reichstein-Vistoli and Chernousov-Merkurjev, strict reductive envelopes of groups of type $A$ by Cernele-Reichstein, and semisimple groups of type $B$ by the authors to any classical type and type $E_{6}$ in a uniform way.