arXiv subjects

Hyemi Kim

Publications and source records attributed to Hyemi Kim.

7 recordsLinked to original sources

arXiv2026

Learning Fair Demand Models

Data-driven pricing is increasingly prevalent in sectors such as airlines, lending, insurance, and retail. By learning demand models from customer features and setting prices accordingly, these systems may generate discriminatory outcomes that raise fairness concerns. This leads to fundamental questions - how and where should systems incorporate fairness considerations in the pricing pipeline, and how does it ultimately affect societal outcomes? To answer these, we study a stylized model where a seller has a two-stage decision pipeline comprising linear demand model estimation followed by price optimization. The seller considers fairness notions in training loss, price, and demand, under both parity-wise and Rawlsian perspectives. We show that equalizing training loss across consumer groups leads to multiple solutions, which in turn can result in undesirable outcomes despite being a standard approach in fair machine learning. Focusing instead on fairness applied directly to prices or demand, we compare two strategies that enforce fairness in either the demand estimation stage or the price optimization stage. For parity-wise fairness, we characterize when each strategy yields higher social welfare under small fairness levels. We show that when market sizes and prices in the dataset are similar, imposing price fairness in the estimation stage is more beneficial to consumers, whereas imposing demand fairness in the optimization stage yields better consumer outcomes. For Rawlsian fairness, the two strategies coincide exactly. Lastly, we extend our model to alternate demand functions and conduct a case study using real-world vaccine pricing data.

cs.CY↗

arXiv2026

Fair Aggregation in Virtual Power Plants

A virtual power plant (VPP) is operated by an aggregator that acts as a market intermediary, aggregating consumers to participate in wholesale power markets. By setting incentive prices, the aggregator induces consumers to sell energy and profits by providing this aggregated energy to the market. This supply is enabled by consumers' flexibility to adjust electricity consumption in response to market conditions. However, heterogeneity in flexibility means that profit-maximizing VPP pricing can create inequalities in participation and benefit allocation across consumers. In this paper, we develop a fairness-aware pricing framework to analyze how different fairness notions reshape system performance, measured by consumer Nash welfare, total consumer utility, and social welfare. We consider three fairness criteria: energy fairness, which ensures equitable energy provision; price fairness, which ensures similar incentive prices; and utility fairness, which ensures comparable levels of consumer utility. We model the aggregator-consumer interaction as a Stackelberg game and derive consumers' optimal responses to incentive prices. Using a stylized model, we show that profit-only pricing systematically disadvantages less flexible consumers. We further show that energy fairness can either improve or worsen all performance measures, and gains across most measures arise only at moderate fairness levels. Surprisingly, price fairness never benefits less flexible consumers, even when it reduces price disparities. By contrast, utility fairness protects less flexible consumers without benefiting more flexible ones. We validate our findings using data from an experiment in Norway under a tiered pricing scheme. Our results provide regulators and VPP operators with a systematic map linking fairness definitions and enforcement levels to operational and welfare outcomes.

cs.GT↗

arXiv2026

Shallow Trap States Control Electrical Performance of Amorphous Oxide Semiconductor Thin-Film Transistors

The performance of n-type amorphous oxide semiconductor thin-film transistors (TFTs) is largely controlled by the density of states (DoS) near the conduction band mobility edge. Here, the full subgap DoS of amorphous InGaZnO (a-IGZO) TFTs, used in display panels and dynamic random-access memory (DRAM) development, is measured by ultrabroadband photoconduction (UP-DoS) microscopy to within 0.1 eV of the mobility edge. The measured subgap DoS for 25 TFT processing conditions accurately predicts each transfer curve, showing how shallow defect states are electron traps that rigidly tune subthreshold swing, threshold voltage and drift mobility. For a set of TFTs, the subthreshold transfer characteristics can be independently simulated from the experimental shallow defect DoS, with no adjustable parameters. The full transfer curve is simulated by introducing a single parameter: the conduction band tail energy. Additionally, the simulation reveals that the shallow trap density controlling subthreshold behavior can be directly extracted from transfer curves. Finally, a systematic In-enrichment study, combined with DFT+U DoS simulations, enables identification of vacancy cation coordination environments for all experimentally observed subgap peaks. The dominant trap controlling conventional a-IGZO TFT performance is centered at ~0.32 eV below the conduction band mobility edge and is assigned to a Ga-Ga-In oxygen vacancy defect.

physics.app-ph↗

arXiv2025

Multi-View Attention Multiple-Instance Learning Enhanced by LLM Reasoning for Cognitive Distortion Detection

Cognitive distortions have been closely linked to mental health disorders, yet their automatic detection remains challenging due to contextual ambiguity, co-occurrence, and semantic overlap. We propose a novel framework that combines Large Language Models (LLMs) with a Multiple-Instance Learning (MIL) architecture to enhance interpretability and expression-level reasoning. Each utterance is decomposed into Emotion, Logic, and Behavior (ELB) components, which are processed by LLMs to infer multiple distortion instances, each with a predicted type, expression, and model-assigned salience score. These instances are integrated via a Multi-View Gated Attention mechanism for final classification. Experiments on Korean (KoACD) and English (Therapist QA) datasets demonstrate that incorporating ELB and LLM-inferred salience scores improves classification performance, especially for distortions with high interpretive ambiguity. Our results suggest a psychologically grounded and generalizable approach for fine-grained reasoning in mental health NLP. The dataset and implementation details are publicly accessible.

cs.CL↗

arXiv2023

A study of audio mixing methods for piano transcription in violin-piano ensembles

While piano music transcription models have shown high performance for solo piano recordings, their performance degrades when applied to ensemble recordings. This study aims to analyze the impact of different data augmentation methods on piano transcription performance, specifically focusing on mixing techniques applied to violin-piano ensembles. We apply mixing methods that consider both harmonic and temporal characteristics of the audio. To create datasets for this study, we generated the PFVN-synth dataset, which contains 7 hours of violin-piano ensemble audio by rendering MIDI files and corresponding labels, and also collected unaccompanied violin recordings and mixed them with the MAESTRO dataset. We evaluated the transcription results on both synthesized and real audio recordings datasets.

cs.SD↗

arXiv2020

Counterfactual Fairness with Disentangled Causal Effect Variational Autoencoder

The problem of fair classification can be mollified if we develop a method to remove the embedded sensitive information from the classification features. This line of separating the sensitive information is developed through the causal inference, and the causal inference enables the counterfactual generations to contrast the what-if case of the opposite sensitive attribute. Along with this separation with the causality, a frequent assumption in the deep latent causal model defines a single latent variable to absorb the entire exogenous uncertainty of the causal graph. However, we claim that such structure cannot distinguish the 1) information caused by the intervention (i.e., sensitive variable) and 2) information correlated with the intervention from the data. Therefore, this paper proposes Disentangled Causal Effect Variational Autoencoder (DCEVAE) to resolve this limitation by disentangling the exogenous uncertainty into two latent variables: either 1) independent to interventions or 2) correlated to interventions without causality. Particularly, our disentangling approach preserves the latent variable correlated to interventions in generating counterfactual examples. We show that our method estimates the total effect and the counterfactual effect without a complete causal graph. By adding a fairness regularization, DCEVAE generates a counterfactual fair dataset while losing less original information. Also, DCEVAE generates natural counterfactual images by only flipping sensitive information. Additionally, we theoretically show the differences in the covariance structures of DCEVAE and prior works from the perspective of the latent disentanglement.

cs.LG↗

arXiv2020

Neutralizing Gender Bias in Word Embedding with Latent Disentanglement and Counterfactual Generation

Recent research demonstrates that word embeddings, trained on the human-generated corpus, have strong gender biases in embedding spaces, and these biases can result in the discriminative results from the various downstream tasks. Whereas the previous methods project word embeddings into a linear subspace for debiasing, we introduce a \textit{Latent Disentanglement} method with a siamese auto-encoder structure with an adapted gradient reversal layer. Our structure enables the separation of the semantic latent information and gender latent information of given word into the disjoint latent dimensions. Afterwards, we introduce a \textit{Counterfactual Generation} to convert the gender information of words, so the original and the modified embeddings can produce a gender-neutralized word embedding after geometric alignment regularization, without loss of semantic information. From the various quantitative and qualitative debiasing experiments, our method shows to be better than existing debiasing methods in debiasing word embeddings. In addition, Our method shows the ability to preserve semantic information during debiasing by minimizing the semantic information losses for extrinsic NLP downstream tasks.

cs.CL↗