arXiv subjects

Yi-Hsin Chen

Publications and source records attributed to Yi-Hsin Chen.

At least 37 records · Page 2Linked to original sources

arXiv2023

Transformer-based Variable-rate Image Compression with Region-of-interest Control

This paper proposes a transformer-based learned image compression system. It is capable of achieving variable-rate compression with a single model while supporting the region-of-interest (ROI) functionality. Inspired by prompt tuning, we introduce prompt generation networks to condition the transformer-based autoencoder of compression. Our prompt generation networks generate content-adaptive tokens according to the input image, an ROI mask, and a rate parameter. The separation of the ROI mask and the rate parameter allows an intuitive way to achieve variable-rate and ROI coding simultaneously. Extensive experiments validate the effectiveness of our proposed method and confirm its superiority over the other competing methods.

eess.IV↗

arXiv2023

Optimizing Population Accumulation in Quantum States Using Microwave Spectroscopy

We present an all-optical method for efficiently preparing cold atoms in a desired Zeeman state, either on the magnetically insensitive clock state (m_F=0) or a particular state suitable for processing or storing quantum information. By applying the theoretical fitting model to a single microwave spectrum, we can individually determine the population distribution, microwave polarization ratio, and microwave Rabi frequency. We can dynamically track the population distribution during the optical pumping process using this real-time microwave spectrum. In a steady-state condition, a simplified model, which considers resonant and off-resonant transitions, indicates that there is an upper limit to the purity under a weak optical pumping field. The population purity up to 96(2)% or 98(1)% on the desired quantum state has been achieved after optimizing the intensity and polarization of the optical pumping field. Our study provides valuable information and potential applications in precision measurement and quantum computation research.

physics.atom-ph↗

arXiv2023

Content-Adaptive Motion Rate Adaption for Learned Video Compression

This paper introduces an online motion rate adaptation scheme for learned video compression, with the aim of achieving content-adaptive coding on individual test sequences to mitigate the domain gap between training and test data. It features a patch-level bit allocation map, termed the $α$-map, to trade off between the bit rates for motion and inter-frame coding in a spatially-adaptive manner. We optimize the $α$-map through an online back-propagation scheme at inference time. Moreover, we incorporate a look-ahead mechanism to consider its impact on future frames. Extensive experimental results confirm that the proposed scheme, when integrated into a conditional learned video codec, is able to adapt motion bit rate effectively, showing much improved rate-distortion performance particularly on test sequences with complicated motion characteristics.

eess.IV↗

arXiv2022

Optical pumping effects on the Rydberg EIT spectrum

We provide a universal discussion of the interplay between Rydberg-state electromagnetically induced transparency (EIT) and optical pumping (OP) in a thermal Rb87 medium. By pumping the population to one single Hyperfine/Zeeman state, we can enhance the interaction strength and, in principle, amplify the EIT peak. According to our measurements, the EIT peak height can be improved by a factor of two or reduced by one order of magnitude, and linewidth was slightly narrowed by the pumping effect. Similar behavior is also seen by increasing the optical density (OD) of the medium. The EIT feature is predicted quantitatively using a Doppler-free non-perturbation numerical calculation. With and without the optical pumping field, the EIT peak heights collapse onto the same theoretical curve, showing that OP and varying OD have the same effect. In both simulations and measurements, Rydberg EIT enhancement through OP is dependent on the intensity of the probe field and the OD. Our work clarifies the underlying mechanisms of optical pumping and advances Rydberg-atom research.

physics.atom-ph↗

arXiv2022

Room-temperature biphoton source with a spectral brightness near the ultimate limit

The biphotons, generated from a hot atomic vapor via the process of spontaneous four-wave mixing (SFWM), have the following merits: stable and tunable frequencies as well as linewidth. Such merits are very useful in the applications of long-distance quantum communication. However, the hot-atom SFWM biphoton sources previously had far lower values of generation rate per linewidth, i.e., spectral brightness, as compared with the sources of biphotons generated by the spontaneous parametric down conversion (SPDC) process. Here, we report a hot-atom SFWM source of biphotons with a linewidth of 960 kHz and a generation rate of 3.7$\times$ $10^5$ pairs/s. The high generation rate, together with the narrow linewidth, results in a spectral brightness of 3.8$\times$ $10^5$ pairs/s/MHz, which is 17 times of the previous best result with atomic vapors and also better than all known results with all kinds of media. The all-copropagating scheme together with a large optical depth (OD) of the atomic vapor is the key improvement, enabling the achieved spectral brightness to be about one quarter of the ultimate limit. Furthermore, this biphoton source had a signal-to-background ratio (SBR) of 2.7, which violated the Cauchy-Schwartz inequality for classical light by about 3.6 folds. Although an increasing spectral brightness usually leads to a decreasing SBR, our systematic study indicates that both of the present spectral brightness and SBR can be enhanced by further increasing the OD. This work demonstrates a significant advancement and provides useful knowledge in the quantum technology using photons.

quant-ph↗

arXiv2021

Optimizing the Rydberg EIT spectrum in a thermal vapor

We present Rydberg-state electromagnetically-induced-transparency (EIT) measurements examining the effects of laser polarization, magnetic fields, laser intensities, and the optical density of the thermal $^{87}$Rb medium. Two counter-propagating laser beams with wavelengths of 480 nm and 780 nm were employed to sweep the spectrum across the Rydberg states $|33D_{3/2}\rangle$ and $|33D_{5/2}\rangle$. An analytic transmission expression well fits the Rydberg-EIT spectra with multiple transitions under different magnetic fields and laser polarization after accounting for the relevant Clebsch-Gordan coefficients, Zeeman splittings, and Doppler shifts. In addition, the high-contrast Rydberg EIT can be optimized with the probe laser intensity and optical density. Rydberg EIT peak height was achieved at $13\%$, which is more than twice as high as the maximum peak height at room temperature. A quantitative theoretical model is employed to represent the spectra properties and to predict well the optimization conditions. A Rydberg EIT spectrum with high contrast in real-time can be served as a quantum sensor to detect the electromagnetic field within an environment.

physics.atom-ph↗

arXiv2021

Video Rescaling Networks with Joint Optimization Strategies for Downscaling and Upscaling

This paper addresses the video rescaling task, which arises from the needs of adapting the video spatial resolution to suit individual viewing devices. We aim to jointly optimize video downscaling and upscaling as a combined task. Most recent studies focus on image-based solutions, which do not consider temporal information. We present two joint optimization approaches based on invertible neural networks with coupling layers. Our Long Short-Term Memory Video Rescaling Network (LSTM-VRN) leverages temporal information in the low-resolution video to form an explicit prediction of the missing high-frequency information for upscaling. Our Multi-input Multi-output Video Rescaling Network (MIMO-VRN) proposes a new strategy for downscaling and upscaling a group of video frames simultaneously. Not only do they outperform the image-based invertible model in terms of quantitative and qualitative results, but also show much improved upscaling quality than the video rescaling methods without joint optimization. To our best knowledge, this work is the first attempt at the joint optimization of video downscaling and upscaling.

cs.CV↗

arXiv2021

Generation of sub-MHz and spectrally-bright biphotons from hot atomic vapors with a phase mismatch-free scheme

We utilized the all-copropagating scheme, which maintains the phase-match condition, in the spontaneous four-wave mixing (SFWM) process to generate biphotons from a hot atomic vapor. The scheme enables our biphotons not only to surpass those in the previous works of hot-atom SFWM, but also to compete with the biphotons that are generated by either the cold-atom SFWM or the cavity-assisted spontaneous parametric down conversion. The biphoton linewidth in this work is tunable for an order of magnitude. As we tuned the linewidth to 610 kHz, the maximum two-photon correlation function, $g_{s,as}^{(2)}$, of the biphotons is 42. This $g_{s,as}^{(2)}$ violates the Cauchy-Schwartz inequality for classical light by 440 folds, and demonstrates that the biphotons have a high purity. The generation rate per linewidth of the 610-kHz biphoton source is 1,500 pairs/(s$\cdot$MHz), which is the best result of all the sub-MHz biphoton sources in the literature. By increasing the pump power by 16 folds, we further enhanced the generation rate per linewidth to 2.3$\times$10$^4$ pairs/(s$\cdot$MHz), while the maximum $g_{s,as}^{(2)}$ became 6.7. In addition, we are able to tune the linewidth down to 290$\pm$20 kHz. This is the narrowest linewidth to date, among all the various kinds of single-mode biphotons.

quant-ph↗

arXiv2021

Class-incremental Learning with Rectified Feature-Graph Preservation

In this paper, we address the problem of distillation-based class-incremental learning with a single head. A central theme of this task is to learn new classes that arrive in sequential phases over time while keeping the model's capability of recognizing seen classes with only limited memory for preserving seen data samples. Many regularization strategies have been proposed to mitigate the phenomenon of catastrophic forgetting. To understand better the essence of these regularizations, we introduce a feature-graph preservation perspective. Insights into their merits and faults motivate our weighted-Euclidean regularization for old knowledge preservation. We further propose rectified cosine normalization and show how it can work with binary cross-entropy to increase class separation for effective learning of new classes. Experimental results on both CIFAR-100 and ImageNet datasets demonstrate that our method outperforms the state-of-the-art approaches in reducing classification error, easing catastrophic forgetting, and encouraging evenly balanced accuracy over different classes. Our project page is at : https://github.com/yhchen12101/FGP-ICL.

cs.CV↗

arXiv2019

EmotionX-IDEA: Emotion BERT -- an Affectional Model for Conversation

In this paper, we investigate the emotion recognition ability of the pre-training language model, namely BERT. By the nature of the framework of BERT, a two-sentence structure, we adapt BERT to continues dialogue emotion prediction tasks, which rely heavily on the sentence-level context-aware understanding. The experiments show that by mapping the continues dialogue into a causal utterance pair, which is constructed by the utterance and the reply utterance, models can better capture the emotions of the reply utterance. The present method has achieved 0.815 and 0.885 micro F1 score in the testing dataset of Friends and EmotionPush, respectively.

cs.CL↗

arXiv2019

Low-Loss High-Fidelity Frequency-Mode Hadamard Gates Based on Electromagnetically Induced Transparency

A frequency beam splitter (FBS) with the split ratio of 0.5 or 1 can be used as the frequency-mode Hadamard gate (FHG) for frequency-encoded photonic qubits or as the quantum frequency converter (QFC) for frequency up or down conversion of photons. Previous works revealed that all kinds of the FHG or QFC operating at the single-photon level had overall efficiency or output-to-input ratio around 50% or less. In this work, our FHG and QFC are made with the four-wave mixing process based on the dual-$Λ$ electromagnetically induced transparency scheme. We achieved an overall efficiency of 90$\pm$4% in the FGH and that of 84% in the QFC using coherent-state single photons, both of which are the best up-to-date records. To test the fidelity of the FBS, we propose a novel scheme of Hong-Ou-Mandel interference (HOMI) for quantum process tomography. The fidelity indicated by the HOMI's $g^{(2)}$ measurement of the FHG is 0.99$\pm$0.01. Such low-loss high-fidelity FHG and QFC or FBS with the tunable split ratio can lead to useful operations or devices in long-distance quantum communication.

quant-ph↗

arXiv2019

Leveraging Linguistic Characteristics for Bipolar Disorder Recognition with Gender Differences

Most previous studies on automatic recognition model for bipolar disorder (BD) were based on both social media and linguistic features. The present study investigates the possibility of adopting only language-based features, namely the syntax and morpheme collocation. We also examine the effect of gender on the results considering gender has long been recognized as an important modulating factor for mental disorders, yet it received little attention in previous linguistic models. The present study collects Twitter posts 3 months prior to the self-disclosure by 349 BD users (231 female, 118 male). We construct a set of syntactic patterns in terms of the word usage based on graph pattern construction and pattern attention mechanism. The factors examined are gender differences, syntactic patterns, and bipolar recognition performance. The performance indicates our F1 scores reach over 91% and outperform several baselines, including those using TF-IDF, LIWC and pre-trained language models (ELMO and BERT). The contributions of the present study are: (1) The features are contextualized, domain-agnostic, and purely linguistic. (2) The performance of BD recognition is improved by gender-enriched linguistic pattern features, which are constructed with gender differences in language usage.

cs.IR↗

arXiv2018

Highly efficient coherent optical memory based on electromagnetically induced transparency

Quantum memory is an important component in the long-distance quantum communication system based on the quantum repeater protocol. To outperform the direct transmission of photons with quantum repeaters, it is crucial to develop quantum memories with high fidelity, high efficiency and a long storage time. Here, we achieve a storage efficiency of 92.0(1.5)\% for a coherent optical memory based on the electromagnetically induced transparency (EIT) scheme in optically dense cold atomic media. We also obtain a useful time-bandwidth product of 1200, considering only storage where the retrieval efficiency remains above 50\%. Both are the best record to date in all kinds of the schemes for the realization of optical memory. Our work significantly advances the pursuit of a high-performance optical memory and should have important applications in quantum information science.

quant-ph↗

arXiv2017

No More Discrimination: Cross City Adaptation of Road Scene Segmenters

Despite the recent success of deep-learning based semantic segmentation, deploying a pre-trained road scene segmenter to a city whose images are not presented in the training set would not achieve satisfactory performance due to dataset biases. Instead of collecting a large number of annotated images of each city of interest to train or refine the segmenter, we propose an unsupervised learning approach to adapt road scene segmenters across different cities. By utilizing Google Street View and its time-machine feature, we can collect unannotated images for each road scene at different times, so that the associated static-object priors can be extracted accordingly. By advancing a joint global and class-specific domain adversarial learning framework, adaptation of pre-trained segmenters to that city can be achieved without the need of any user annotation or interaction. We show that our method improves the performance of semantic segmentation in multiple cities across continents, while it performs favorably against state-of-the-art approaches requiring annotated training data.

cs.CV↗

arXiv2017

Enhanced spectral profile in the study of Doppler-broadened Rydberg atoms

Combination of the electromagnetically-induced-transparency (EIT) effect and Rydberg-state atoms has attracted great attention recently due to its potential application in the photon-photon interaction or qubit operation. In this work, we studied the Rydberg-EIT spectra with room-temperature $^{87}$Rb atoms. Spectroscopic data under various experimental parameters all showed that the contrast of EIT transparency as a function of the probe intensity is initially enhanced, reaches a maximum value and then decays gradually. The contrast of spectral profile at the optimum probe field intensity is enhanced by $2-4$ times as compared with that at weak intensity. Moreover, the signal-to-noise ratio of the spectrum can potentially be improved by 1 or 2 order of magnitude. We provided a theoretical model to explain this behavior and clarified its underlying mechanism. Our work overcomes the obstacle of weak signal in the Rydberg-EIT spectrum caused by an apparent relaxation rate of the Rydberg polariton and weak coupling transition strength, and provides the useful knowledge for the Rydberg-EIT study.

physics.atom-ph↗

arXiv2016

Large Cross-Phase Modulations at the Few-Photon Level

We demonstrate an efficient cross-phase modulation (XPM) based on a closed-loop double-Λ system. The property of the double-Λ medium can be controlled by changing the phases of the applied optical fields. This phase-dependent XPM scheme can achieve large phase modulations at low-light intensities without requiring cavities or tightly focusing of laser beams. With this scheme, we observe a π-level phase shift with two pulses both consisting of 8 photons in cold rubidium atoms. Such novel scheme provides a simple route to generate strong interactions between photons and may have potential applications in all-optical quantum signal processing.

quant-ph↗

arXiv2016

Rydberg polaritons in a thermal vapor

We present a pulsed four-wave mixing (FWM) scheme via a Rydberg state to create, store and retrieve collective Rydberg polaritons. The storage medium consists of a gas of thermal Rb atoms confined in a 220 μm thick cell, which are heated above room temperature. The experimental sequence consists of a pulsed excitation of Rydberg polaritons via the D1 line, a variable delay or storage time, and a final retrieval pulse via the D2 line. The lifetime of the Rydberg polaritons is around 1.2 ns, almost entirely limited by the excitation bandwidth and the corresponding motional dephasing of the atoms. The presented scheme combined with a tightly confined atomic ensemble is a good candidate for a deterministic single-photon source, as soon as strong interactions in terms of a Rydberg blockade are added.

physics.atom-ph↗

arXiv2015

Pulsed Rydberg four-wave mixing with motion-induced dephasing in a thermal vapor

We report on time-resolved pulsed four-wave mixing (FWM) signals in a thermal Rubidium vapor involving a Rydberg state. We observe FWM signals with dephasing times up to 7 ns, strongly dependent on the excitation bandwidth to the Rydberg state. The excitation to the Rydberg state is driven by a pulsed two-photon transition on ns time scales. Combined with a third cw de-excitation laser, a strongly directional and collective emission is generated according to a combination of the phase matching effect and averaging over Doppler classes. In contrast to a previous report [1] using off-resonant FWM, at a resonant FWM scheme we observe additional revivals of the signal shortly after the incident pulse has ended. We infer that this is a revival of motion-induced constructive interference between the coherent emissions of the thermal atoms. The resonant FWM scheme reveals a richer temporal structure of the signals, compared to similar, but off-resonant excitation schemes. A simple explanation lies in the selectivity of Doppler classes. Our numerical simulations based on a four-level model including a whole Doppler ensemble can qualitatively describe the data.

physics.atom-ph↗