arXiv subjects

Yi-Chun Chen

Publications and source records attributed to Yi-Chun Chen.

34 records · Page 2Linked to original sources

arXiv2023

Panel Transitions for Genre Analysis in Visual Narratives

Understanding how humans communicate and perceive narratives is important for media technology research and development. This is particularly important in current times when there are tools and algorithms that are easily available for amateur users to create high-quality content. Narrative media develops over time a set of recognizable patterns of features across similar artifacts. Genre is one such grouping of artifacts for narrative media with similar patterns, tropes, and story structures. While much work has been done on genre-based classifications in text and video, we present a novel approach to do a multi-modal analysis of genre based on comics and manga-style visual narratives. We present a systematic feature analysis of an annotated dataset that includes a variety of western and eastern visual books with annotations for high-level narrative patterns. We then present a detailed analysis of the contributions of high-level features to genre classification for this medium. We highlight some of the limitations and challenges of our existing computational approaches in modeling subjective labels. Our contributions to the community are: a dataset of annotated manga books, a multi-modal analysis of visual panels and text in a constrained and popular medium through high-level features, and a systematic process for incorporating subjective narrative patterns in computational models.

cs.AI↗

arXiv2023

A Customizable Generator for Comic-Style Visual Narrative

We present a theory-inspired visual narrative generator that incorporates comic-authoring idioms, which transfers the conceptual principles of comics into system layers that integrate the theories to create comic content. The generator creates comics through sequential decision-making across layers from panel composition, object positions, panel transitions, and narrative elements. Each layer's decisions are based on narrative goals and follow the respective layer idioms of the medium. Cohn's narrative grammar provides the overall story arc. Photographic compositions inspired by the rule of thirds is used to provide panel compositions. McCloud's proposed panel transitions based on focus shifts between scene, character, and temporal changes are encoded in the transition layer. Finally, common overlay symbols (such as the exclamation) are added based on analyzing action verbs using an action-verb ontology. We demonstrate the variety of generated comics through various settings with example outputs. The generator and associated modules could be a useful system for visual narrative authoring and for further research into computational models of visual narrative understanding.

cs.AI↗

arXiv2023

Direct Implementation with Evidence

We study full implementation with evidence in an environment with bounded utilities. We show that a social choice function is Nash implementable in a direct revelation mechanism if and only if it satisfies the measurability condition proposed by BL2012 . Building on a novel classification of lies according to their refutability with evidence, the mechanism requires only two agents, accounts for mixed-strategy equilibria and accommodates evidentiary costs. While monetary transfers are used, they are off the equilibrium and can be balanced with three or more agents. In a richer model of evidence due to KT2012 , we establish pure-strategy implementation with two or more agents in a direct revelation mechanism. We also obtain a necessary and sufficient condition on the evidence structure for renegotiation-proof bilateral contracts, based on the classification of lies.

econ.TH↗

arXiv2022

Information Design in Allocation with Costly Verification

A principal who values an object allocates it to one or more agents. Agents learn private information (signals) from an information designer about the allocation payoff to the principal. Monetary transfer is not available but the principal can costly verify agents' private signals. The information designer can influence the agents' signal distributions, based upon which the principal maximizes the allocation surplus. An agent's utility is simply the probability of obtaining the good. With a single agent, we characterize (i) the agent-optimal information, (ii) the principal-worst information, and (iii) the principal-optimal information. Even though the objectives of the principal and the agent are not directly comparable, we find that any agent-optimal information is principal-worst. Moreover, there exists a robust mechanism that achieves the principal's payoff under (ii), which is therefore an optimal robust mechanism. Many of our results extend to the multiple-agent case; if not, we provide counterexamples.

econ.TH↗

arXiv2022

Information Design in Optimal Auctions

We study the information design problem in a single-unit auction setting. The information designer controls independent private signals according to which the buyers infer their binary private values. Assuming that the seller adopts the optimal auction due to Myerson (1981) in response, we characterize both the buyer-optimal information structure, which maximizes the buyers' surplus, and the sellerworst information structure, which minimizes the seller's revenue. We translate both information design problems into finite-dimensional, constrained optimization problems in which one can explicitly solve for the optimal information structures. In contrast to the case with one buyer (Roesler and Szentes, 2017), we show that with two or more buyers, the symmetric buyer-optimal information structure is different from the symmetric seller-worst information structure. The good is always sold under the seller-worst information structure but not under the buyer-optimal information structure. Nevertheless, as the number of buyers goes to infinity, both symmetric information structures converge to no disclosure. We also show that in an ex ante symmetric setting, an asymmetric information structure is never seller-worst but can generate a strictly higher surplus for the buyers than the symmetric buyer-optimal information structure.

econ.TH↗

arXiv2022

Maskin Meets Abreu and Matsushima

The theory of full implementation has been criticized for using integer/modulo games which admit no equilibrium (Jackson (1992)). To address the critique, we revisit the classical Nash implementation problem due to Maskin (1999) but allow for the use of lotteries and monetary transfers as in Abreu and Matsushima (1992, 1994). We unify the two well-established but somewhat orthogonal approaches in full implementation theory. We show that Maskin monotonicity is a necessary and sufficient condition for (exact) mixed-strategy Nash implementation by a finite mechanism. In contrast to previous papers, our approach possesses the following features: finite mechanisms (with no integer or modulo game) are used; mixed strategies are handled explicitly; neither undesirable outcomes nor transfers occur in equilibrium; the size of transfers can be made arbitrarily small; and our mechanism is robust to information perturbations. Finally, our result can be extended to infinite/continuous settings and ordinal settings.

econ.TH↗

arXiv2021

Decision Forest: A Nonparametric Approach to Modeling Irrational Choice

Customer behavior is often assumed to follow weak rationality, which implies that adding a product to an assortment will not increase the choice probability of another product in that assortment. However, an increasing amount of research has revealed that customers are not necessarily rational when making decisions. In this paper, we propose a new nonparametric choice model that relaxes this assumption and can model a wider range of customer behavior, such as decoy effects between products. In this model, each customer type is associated with a binary decision tree, which represents a decision process for making a purchase based on checking for the existence of specific products in the assortment. Together with a probability distribution over customer types, we show that the resulting model -- a decision forest -- is able to represent any customer choice model, including models that are inconsistent with weak rationality. We theoretically characterize the depth of the forest needed to fit a data set of historical assortments and prove that with high probability, a forest whose depth scales logarithmically in the number of assortments is sufficient to fit most data sets. We also propose two practical algorithms -- one based on column generation and one based on random sampling -- for estimating such models from data. Using synthetic data and real transaction data exhibiting non-rational behavior, we show that the model outperforms both rational and non-rational benchmark models in out-of-sample predictive ability.

cs.LG↗

arXiv2021

Investigating Ground-level Ozone Formation: A Case Study in Taiwan

Tropospheric ozone (O3) is a greenhouse gas which can absorb heat and make the weather even hotter during extreme heatwaves. Besides, it is an influential ground-level air pollutant which can severely damage the environment. Thus evaluating the importance of various factors related to the O3 formation process is essential. However, O3 simulated by the available climate models exhibits large variance in different places, indicating the insufficiency of models in explaining the O3 formation process correctly. In this paper, we aim to identify and understand the impact of various factors on O3 formation and predict the O3 concentrations under different pollution-reduced and climate change scenarios. We employ six supervised methods to estimate the observed O3 using fourteen meteorological and chemical variables. We find that the deep neural network (DNN) and long short-term memory (LSTM) based models can predict O3 concentrations accurately. We also demonstrate the importance of several variables in this prediction task. The results suggest that while Nitrogen Oxides negatively contributes to predicting O3, solar radiation makes a significantly positive contribution. Furthermore, we apply our two best models on O3 prediction under different global warming and pollution reduction scenarios to improve the policy-making decisions in the O3 reduction.

physics.ao-ph↗

arXiv2020

TrustMAE: A Noise-Resilient Defect Classification Framework using Memory-Augmented Auto-Encoders with Trust Regions

In this paper, we propose a framework called TrustMAE to address the problem of product defect classification. Instead of relying on defective images that are difficult to collect and laborious to label, our framework can accept datasets with unlabeled images. Moreover, unlike most anomaly detection methods, our approach is robust against noises, or defective images, in the training dataset. Our framework uses a memory-augmented auto-encoder with a sparse memory addressing scheme to avoid over-generalizing the auto-encoder, and a novel trust-region memory updating scheme to keep the noises away from the memory slots. The result is a framework that can reconstruct defect-free images and identify the defective regions using a perceptual distance network. When compared against various state-of-the-art baselines, our approach performs competitively under noise-free MVTec datasets. More importantly, it remains effective at a noise level up to 40% while significantly outperforming other baselines.

cs.CV↗

arXiv2019

Deep Reinforcement Learning for Event-Driven Multi-Agent Decision Processes

The incorporation of macro-actions (temporally extended actions) into multi-agent decision problems has the potential to address the curse of dimensionality associated with such decision problems. Since macro-actions last for stochastic durations, multiple agents executing decentralized policies in cooperative environments must act asynchronously. We present an algorithm that modifies generalized advantage estimation for temporally extended actions, allowing a state-of-the-art policy optimization algorithm to optimize policies in Dec-POMDPs in which agents act asynchronously. We show that our algorithm is capable of learning optimal policies in two cooperative domains, one involving real-time bus holding control and one involving wildfire fighting with unmanned aircraft. Our algorithm works by framing problems as "event-driven decision processes," which are scenarios in which the sequence and timing of actions and events are random and governed by an underlying stochastic process. In addition to optimizing policies with continuous state and action spaces, our algorithm also facilitates the use of event-driven simulators, which do not require time to be discretized into time-steps. We demonstrate the benefit of using event-driven simulation in the context of multiple agents taking asynchronous actions. We show that fixed time-step simulation risks obfuscating the sequence in which closely separated events occur, adversely affecting the policies learned. In addition, we show that arbitrarily shrinking the time-step scales poorly with the number of agents.

cs.AI↗

arXiv2018

Learning Discrete Bayesian Networks from Continuous Data

Learning Bayesian networks from raw data can help provide insights into the relationships between variables. While real data often contains a mixture of discrete and continuous-valued variables, many Bayesian network structure learning algorithms assume all random variables are discrete. Thus, continuous variables are often discretized when learning a Bayesian network. However, the choice of discretization policy has significant impact on the accuracy, speed, and interpretability of the resulting models. This paper introduces a principled Bayesian discretization method for continuous variables in Bayesian networks with quadratic complexity instead of the cubic complexity of other standard techniques. Empirical demonstrations show that the proposed method is superior to the established minimum description length algorithm. In addition, this paper shows how to incorporate existing methods into the structure learning process to discretize all continuous variables and simultaneously learn Bayesian network structures.

cs.AI↗

arXiv2017

Self-view Grounding Given a Narrated 360° Video

Narrated 360° videos are typically provided in many touring scenarios to mimic real-world experience. However, previous work has shown that smart assistance (i.e., providing visual guidance) can significantly help users to follow the Normal Field of View (NFoV) corresponding to the narrative. In this project, we aim at automatically grounding the NFoVs of a 360° video given subtitles of the narrative (referred to as "NFoV-grounding"). We propose a novel Visual Grounding Model (VGM) to implicitly and efficiently predict the NFoVs given the video content and subtitles. Specifically, at each frame, we efficiently encode the panorama into feature map of candidate NFoVs using a Convolutional Neural Network (CNN) and the subtitles to the same hidden space using an RNN with Gated Recurrent Units (GRU). Then, we apply soft-attention on candidate NFoVs to trigger sentence decoder aiming to minimize the reconstruct loss between the generated and given sentence. Finally, we obtain the NFoV as the candidate NFoV with the maximum attention without any human supervision. To train VGM more robustly, we also generate a reverse sentence conditioning on one minus the soft-attention such that the attention focuses on candidate NFoVs less relevant to the given sentence. The negative log reconstruction loss of the reverse sentence (referred to as "irrelevant loss") is jointly minimized to encourage the reverse sentence to be different from the given sentence. To evaluate our method, we collect the first narrated 360° videos dataset and achieve state-of-the-art NFoV-grounding performance.

cs.CV↗

arXiv2016

Efficient n-type Doping in Epitaxial Graphene through Strong Lateral Orbital Hybridization of Ti Adsorbate

In recent years, various doping methods for epitaxial graphene have been demonstrated through atom substitution and adsorption. Here we observe by angle-resolved photoemission spectroscopy (ARPES) a coupling-induced Dirac cone renormalization when depositing small amounts of Ti onto epitaxial graphene on SiC. We obtain a remarkably high doping efficiency and a readily tunable carrier velocity simply by changing the amount of deposited Ti. First-principles theoretical calculations show that a strong lateral (non-vertical) orbital coupling leads to an efficient doping of graphene by hybridizing the 2pz orbital of graphene and the 3d orbitals of the Ti adsorbate, which attached on graphene without creating any trap/scattering states. This Ti-induced hybridization is adsorbate-specific and has major consequences for efficient doping as well as applications towards adsorbate-induced modification of carrier transport in graphene.

cond-mat.mes-hall↗

arXiv2013

Tuning the Competition between Ferromagnetism and Antiferromagnetism in a Half-doped Manganite through Magnetoelectric Coupling

We investigate the possibility of controlling the magnetic phase transition of the heterointerface between a half-doped manganite La$_{0.5}$Ca$_{0.5}$MnO$_{3}$ and a multiferroic BiFeO$_{3}$ through magnetoelectric coupling. Using macroscopic magnetometry and element-selective x-ray magnetic circular dichroism at the Mn and Fe \emph{L}-edges, we discover that the ferroelectric polarization of BFO controls simultaneously the magnetization of BFO and LCMO. X-ray absorption spectra at oxygen \emph{K}-edge and linear dichroism at Mn \emph{L}-edge suggest that the interfacial coupling is mainly derived from the superexchange between Mn and Fe \emph{t}$_{2g}$ spins. The combination of x-ray absorption spectroscopy and mean field theory calculations reveals that the \emph{d}-electron modulation of Mn cations changes the magnetic coupling in LCMO, which controls the enhanced canted moments of interfacial BFO via the interfacial coupling. Our results demonstrate that the competition between ferromagnetic and antiferromagnetic instability can be modulated by an electric field at the heterointerface, providing another pathway for the electrical field control of magnetism.

cond-mat.str-el↗

arXiv2012

Local Conduction at the BiFeO3-CoFe2O4 Tubular Oxide Interface

In strongly correlated oxides, heterointerfaces, manipulating the interaction, frustration, and discontinuity of lattice, charge, orbital, and spin degrees of freedom, generate new possibilities for next generation devices. In this study, we went back to examine the existing oxide heterostructures and found the local conduction at the BiFeO3-CoFe2O4 vertical interface. In such hetero-nanostructure, the tubular interface, surrounding BiFeO3-CoFe2O4 vertical interface, can not only be the medium to the coupling between phases, but also be a new state of the matter. Our study demonstrates a novel concept on oxide interface design and opens a pathway alternative for the explorations of diverse functionalities in complex oxide interfaces.

cond-mat.mtrl-sci↗

arXiv2012

Graphene on Au-coated SiOx substrate: Its core-level photoelectron micro-spectroscopy study

The core-level electronic structures of the exfoliated graphene sheets on a Au-coated SiOx substrate have been studied by synchrotron radiation photoelectron spectroscopy (SR-PES) on a micron-scale. The graphene was firstly demonstrated its visibility on the Au-coated SiOx substrate by micro-optical characterization, and then conducted into SR-PES study. Because of the elimination of charging effect, precise C 1s core-level characterization clearly shows graphitic and contaminated carbon states of graphene. Different levels of Au-coating-induced p-type doping on single- and double-layer graphene sheets were also examined in the C 1s core-level shift. The Au-coated SiOx substrate can be treated as a simple but high-throughput platform for in situ studying graphene under further hybridization by PES.

cond-mat.mes-hall↗