Search

Most Americans today need some education beyond high school to secure a well-paying job, and higher education is a key factor in social and economic mobility. But higher education is not accessible to all students. To further equalize opportunity, states must understand options for improving the way they fund higher education. America's public K-12 system has been grappling with similar funding disparities for decades -- challenges that offer important lessons for postsecondary funding. Bellwether's analysis, "Strengthening State Higher Education Funding: Lessons Learned From K-12," identifies insights and lessons from K-12 finance policy that may be instructive as states rework how they fund higher education. This report explores four topics and associated questions that are also relevant across K-12 education: (1) Equity: How can states distribute funding to students and public IHEs with the greatest needs?; (2) Adequacy: How much funding is enough to achieve the desired outcomes?; (3) Cost Sharing: How can local governments and states share the cost of public higher education?; and (4) Stability: How can public IHEs secure consistent and predictable funding?

In the training and inference of spiking neural networks (SNNs), direct training and lightweight computation methods have been orthogonally developed, aimed at reducing power consumption. However, only a limited number of approaches have applied these two mechanisms simultaneously and failed to fully leverage the advantages of SNN-based vision transformers (ViTs) since they were originally designed for convolutional neural networks (CNNs). In this paper, we propose AT-SNN designed to dynamically adjust the number of tokens processed during inference in SNN-based ViTs with direct training, wherein power consumption is proportional to the number of tokens. We first demonstrate the applicability of adaptive computation time (ACT), previously limited to RNNs and ViTs, to SNN-based ViTs, enhancing it to discard less informative spatial tokens selectively. Also, we propose a new token-merge mechanism that relies on the similarity of tokens, which further reduces the number of tokens while enhancing accuracy. We implement AT-SNN to Spikformer and show the effectiveness of AT-SNN in achieving high energy efficiency and accuracy compared to state-of-the-art approaches on the image classification tasks, CIFAR10, CIFAR-100, and TinyImageNet. For example, our approach uses up to 42.4% fewer tokens than the existing best-performing method on CIFAR-100, while conserving higher accuracy., Comment: 8 pages

Exploring the dynamics of an optically levitated dielectric micro- and nanoparticle is an exciting new subject in quantum science. Recent years have witnessed rapid advancements in attaining quantum-limited optical detection and control of a nanoscale particle by coupling its motion to a high-finesse optical cavity in the resolved-sideband regime. In order to control the particle deeper in the quantum regime, however, it is necessary to significantly enhance the coupling between the particle and the cavity. Here, we present a novel platform that can allow for achieving this. Our system consists of a conventional optical tweezer and a toroidal optical microcavity with an ultrahigh quality (Q) factor. The optomechanical coupling between the particle and the cavity is established by placing the particle in the near field of the cavity. The significantly reduced mode volume allows us to achieve a 100-fold increase in the single photon optomechanical coupling compared to a conventional Fabry-P\'erot cavity with macroscopic mirrors, while ultralow loss of the cavity maintains the system close to the resolved-sideband regime. Our approach paves the way for enabling quantum experiments on levitated mesoscopic particles with high quantum cooperativity near the resolved-sideband regime.

Propagation-based video inpainting using optical flow at the pixel or feature level has recently garnered significant attention. However, it has limitations such as the inaccuracy of optical flow prediction and the propagation of noise over time. These issues result in non-uniform noise and time consistency problems throughout the video, which are particularly pronounced when the removed area is large and involves substantial movement. To address these issues, we propose a novel First Frame Filling Video Diffusion Inpainting model (FFF-VDI). We design FFF-VDI inspired by the capabilities of pre-trained image-to-video diffusion models that can transform the first frame image into a highly natural video. To apply this to the video inpainting task, we propagate the noise latent information of future frames to fill the masked areas of the first frame's noise latent code. Next, we fine-tune the pre-trained image-to-video diffusion model to generate the inpainted video. The proposed model addresses the limitations of existing methods that rely on optical flow quality, producing much more natural and temporally consistent videos. This proposed approach is the first to effectively integrate image-to-video diffusion models into video inpainting tasks. Through various comparative experiments, we demonstrate that the proposed model can robustly handle diverse inpainting types with high quality.

This study addresses the challenge of fleet design optimization in the context of heterogeneous multi-robot fleets, aiming to obtain feasible designs that balance performance and costs. In the domain of autonomous multi-robot exploration, reinforcement learning agents play a central role, offering adaptability to complex terrains and facilitating collaboration among robots. However, modifying the fleet composition results in changes in the learned behavior, and training multi-robot systems using multi-agent reinforcement learning is expensive. Therefore, an exhaustive evaluation of each potential fleet design is infeasible. To tackle these hurdles, we introduce Bayesian Optimization for Fleet Design (BOFD), a framework leveraging multi-objective Bayesian Optimization to explore fleets on the Pareto front of performance and cost while accounting for uncertainty in the design space. Moreover, we establish a sub-linear bound for cumulative regret, supporting BOFD's robustness and efficacy. Extensive benchmark experiments in synthetic and simulated environments demonstrate the superiority of our framework over state-of-the-art methods, achieving efficient fleet designs with minimal fleet evaluations.

In this work, we discuss evaluating video foundation models in a fair and robust manner. Unlike language or image foundation models, many video foundation models are evaluated with differing parameters (such as sampling rate, number of frames, pretraining steps, etc.), making fair and robust comparisons challenging. Therefore, we present a carefully designed evaluation framework for measuring two core capabilities of video comprehension: appearance and motion understanding. Our findings reveal that existing video foundation models, whether text-supervised like UMT or InternVideo2, or self-supervised like V-JEPA, exhibit limitations in at least one of these capabilities. As an alternative, we introduce TWLV-I, a new video foundation model that constructs robust visual representations for both motion- and appearance-based videos. Based on the average top-1 accuracy of linear probing on five action recognition benchmarks, pretrained only on publicly accessible datasets, our model shows a 4.6%p improvement compared to V-JEPA (ViT-L) and a 7.7%p improvement compared to UMT (ViT-L). Even when compared to much larger models, our model demonstrates a 7.2%p improvement compared to DFN (ViT-H), a 2.7%p improvement compared to V-JEPA (ViT-H) and a 2.8%p improvement compared to InternVideo2 (ViT-g). We provide embedding vectors obtained by TWLV-I from videos of several commonly used video benchmarks, along with evaluation source code that can directly utilize these embeddings. The code is available at https://github.com/twelvelabs-io/video-embeddings-evaluation-framework., Comment: 17 pages; Twelve Labs Technical Report

Building on previous works to construct a homogeneous sample of brown dwarfs in binary systems, we investigate microlensing events detected by the Korea Microlensing Telescope Network (KMTNet) survey during the 2022 and 2023 seasons. Given the difficulty in distinguishing brown-dwarf events from those produced by binary lenses with nearly equal-mass components, we analyze all lensing events detected during the seasons that exhibit anomalies characteristic of binary-lens systems. Using the same criteria consistently applied in previous studies, we identify six additional brown dwarf candidates through the analysis of lensing events KMT-2022-BLG-0412, KMT-2022-BLG-2286, KMT-2023-BLG-0201, KMT-2023-BLG-0601, KMT-2023-BLG-1684, and KMT-2023-BLG-1743. An examination of the mass posteriors shows that the median mass of the lens companions ranges from 0.02 $M_\odot$ to 0.05 $M_\odot$, indicating that these companions fall within the brown-dwarf mass range. The mass of the primary lenses ranges from 0.11 $M_\odot$ to 0.68 $M_\odot$, indicating that they are low-mass stars with substantially lower masses compared to the Sun., Comment: 13 pages, 17 figures, 12 tables

Optical coherence tomography (OCT) has become critical for diagnosing retinal diseases as it enables 3D images of the retina and optic nerve. OCT acquisition is fast, non-invasive, affordable, and scalable. Due to its broad applicability, massive numbers of OCT images have been accumulated in routine exams, making it possible to train large-scale foundation models that can generalize to various diagnostic tasks using OCT images. Nevertheless, existing foundation models for OCT only consider 2D image slices, overlooking the rich 3D structure. Here, we present OCTCube, a 3D foundation model pre-trained on 26,605 3D OCT volumes encompassing 1.62 million 2D OCT images. OCTCube is developed based on 3D masked autoencoders and exploits FlashAttention to reduce the larger GPU memory usage caused by modeling 3D volumes. OCTCube outperforms 2D models when predicting 8 retinal diseases in both inductive and cross-dataset settings, indicating that utilizing the 3D structure in the model instead of 2D data results in significant improvement. OCTCube further shows superior performance on cross-device prediction and when predicting systemic diseases, such as diabetes and hypertension, further demonstrating its strong generalizability. Finally, we propose a contrastive-self-supervised-learning-based OCT-IR pre-training framework (COIP) for cross-modality analysis on OCT and infrared retinal (IR) images, where the OCT volumes are embedded using OCTCube. We demonstrate that COIP enables accurate alignment between OCT and IR en face images. Collectively, OCTCube, a 3D OCT foundation model, demonstrates significantly better performance against 2D models on 27 out of 29 tasks and comparable performance on the other two tasks, paving the way for AI-based retinal disease diagnosis.

The jet cross-section and jet-substructure observables in $p$$+$$p$ collisions at $\sqrt{s}=200$ GeV were measured by the PHENIX Collaboration at the Relativistic Heavy Ion Collider (RHIC). Jets are reconstructed from charged-particle tracks and electromagnetic-calorimeter clusters using the anti-$k_{t}$ algorithm with a jet radius $R=0.3$ for jets with transverse momentum within $8.0

The advent of computational and numerical methods in recent times has provided new avenues for analyzing art historiographical narratives and tracing the evolution of art styles therein. Here, we investigate an evolutionary process underpinning the emergence and stylization of contemporary user-generated visual art styles using the complexity-entropy (C-H) plane, which quantifies local structures in paintings. Informatizing 149,780 images curated in DeviantArt and Behance platforms from 2010 to 2020, we analyze the relationship between local information of the C-H space and multi-level image features generated by a deep neural network and a feature extraction algorithm. The results reveal significant statistical relationships between the C-H information of visual artistic styles and the dissimilarities of the multi-level image features over time within groups of artworks. By disclosing a particular C-H region where the diversity of image representations is noticeably manifested, our analyses reveal an empirical condition of emerging styles that are both novel in the C-H plane and characterized by greater stylistic diversity. Our research shows that visual art analyses combined with physics-inspired methodologies and machine learning, can provide macroscopic insights into quantitatively mapping relevant characteristics of an evolutionary process underpinning the creative stylization of uncharted visual arts of given groups and time., Comment: 18 pages, 3 figures, 1 table, SI(4 figures, 3 tables)

COSINE-100 aims to conclusively test the claimed dark matter annual modulation signal detected by DAMA/LIBRA collaboration. DAMA/LIBRA has released updated analysis results by lowering the energy threshold to 0.75 keV through various upgrades. They have consistently claimed to have observed the annual modulation. In COSINE-100, it is crucial to lower the energy threshold for a direct comparison with DAMA/LIBRA, which also enhances the sensitivity of the search for low-mass dark matter, enabling COSINE-100 to explore this area. Therefore, it is essential to have a precise and quantitative understanding of the background spectrum across all energy ranges. This study expands the background modeling from 0.7 to 4000 keV using 2.82 years of COSINE-100 data. The modeling has been improved to describe the background spectrum across all energy ranges accurately. Assessments of the background spectrum are presented, considering the nonproportionality of NaI(Tl) crystals at both low and high energies and the characteristic X-rays produced by the interaction of external backgrounds with materials such as copper. Additionally, constraints on the fit parameters obtained from the alpha spectrum modeling fit are integrated into this model. These improvements are detailed in the paper.

This paper delves into the emerging field of face-based voice conversion, leveraging the unique relationship between an individual's facial features and their vocal characteristics. We present a novel face-based voice conversion framework that particularly utilizes the average fundamental frequency of the target speaker, derived solely from their facial images. Through extensive analysis, our framework demonstrates superior speech generation quality and the ability to align facial features with voice characteristics, including tracking of the target speaker's fundamental frequency., Comment: Interspeech 2024

When solving forecasting problems including multiple time-series features, existing approaches often fall into two extreme categories, depending on whether to utilize inter-feature information: univariate and complete-multivariate models. Unlike univariate cases which ignore the information, complete-multivariate models compute relationships among a complete set of features. However, despite the potential advantage of leveraging the additional information, complete-multivariate models sometimes underperform univariate ones. Therefore, our research aims to explore a middle ground between these two by introducing what we term Partial-Multivariate models where a neural network captures only partial relationships, that is, dependencies within subsets of all features. To this end, we propose PMformer, a Transformer-based partial-multivariate model, with its training algorithm. We demonstrate that PMformer outperforms various univariate and complete-multivariate models, providing a theoretical rationale and empirical analysis for its superiority. Additionally, by proposing an inference technique for PMformer, the forecasting accuracy is further enhanced. Finally, we highlight other advantages of PMformer: efficiency and robustness under missing features., Comment: 25 pages

Complex physical systems are often described by partial differential equations (PDEs) that depend on parameters such as the Reynolds number in fluid mechanics. In applications such as design optimization or uncertainty quantification, solutions of those PDEs need to be evaluated at numerous points in the parameter space. While physics-informed neural networks (PINNs) have emerged as a new strong competitor as a surrogate, their usage in this scenario remains underexplored due to the inherent need for repetitive and time-consuming training. In this paper, we address this problem by proposing a novel extension, parameterized physics-informed neural networks (P$^2$INNs). P$^2$INNs enable modeling the solutions of parameterized PDEs via explicitly encoding a latent representation of PDE parameters. With the extensive empirical evaluation, we demonstrate that P$^2$INNs outperform the baselines both in accuracy and parameter efficiency on benchmark 1D and 2D parameterized PDEs and are also effective in overcoming the known "failure modes".

We introduce a novel problem setting for algorithmic contract design, named the principal-MARL contract design problem. This setting extends traditional contract design to account for dynamic and stochastic environments using Markov Games and Multi-Agent Reinforcement Learning. To tackle this problem, we propose a Multi-Objective Bayesian Optimization (MOBO) framework named Constrained Pareto Maximum Entropy Search (cPMES). Our approach integrates MOBO and MARL to explore the highly constrained contract design space, identifying promising incentive and recruitment decisions. cPMES transforms the principal-MARL contract design problem into an unconstrained multi-objective problem, leveraging the probability of feasibility as part of the objectives and ensuring promising designs predicted on the feasibility border are included in the Pareto front. By focusing the entropy prediction on designs within the Pareto set, cPMES mitigates the risk of the search strategy being overwhelmed by entropy from constraints. We demonstrate the effectiveness of cPMES through extensive benchmark studies in synthetic and simulated environments, showing its ability to find feasible contract designs that maximize the principal's objectives. Additionally, we provide theoretical support with a sub-linear regret bound concerning the number of iterations.

Identifying the point of error is imperative in software debugging. Traditional fault localization (FL) techniques rely on executing the program and using the code coverage matrix in tandem with test case results to calculate a suspiciousness score for each function or line. Recently, learning-based FL techniques have harnessed machine learning models to extract meaningful features from the code coverage matrix and improve FL performance. These techniques, however, require compilable source code, existing test cases, and specialized tools for generating the code coverage matrix for each programming language of interest. In this paper, we propose, for the first time, a simple but effective sequence generation approach for fine-tuning large language models of code (LLMCs) for FL tasks. LLMCs have recently received much attention for various software engineering problems. In line with these, we leverage the innate understanding of code that LLMCs have acquired through pre-training on large code corpora. Specifically, we fine-tune representative encoder, encoder-decoder, and decoder-based 13 LLMCs for FL tasks. Unlike previous approaches, LLMCs can analyze code sequences even with syntactic errors, since they do not rely on compiled input. Still, they have a limitation on the length of the input data. Therefore, for a fair comparison with existing FL techniques, we extract methods with errors from the project-level benchmark, Defects4J, and analyze them at the line level. Experimental results show that LLMCs fine-tuned with our approach successfully pinpoint error positions in 50.6\%, 64.2\%, and 72.3\% of 1,291 methods in Defects4J for Top-1/3/5 prediction, outperforming the best learning-based state-of-the-art technique by up to 1.35, 1.12, and 1.08 times, respectively. Our findings suggest promising research directions for FL and automated program repair tasks using LLMCs.

We report the first observation of the nearby giant radio galaxy NGC 315 using a global VLBI array consisting of 22 radio antennas located across five continents, including high-sensitivity stations, at 22 GHz. Utilizing the extensive $(u,v)$-coverage provided by the array, coupled with the application of a recently developed super-resolution imaging technique based on the regularized maximum likelihood method, we were able to transversely resolve the NGC 315 jet at parsec scales for the first time. Previously known for its central ridge-brightened morphology at similar scales in former VLBI studies, the jet now clearly exhibits a limb-brightened structure. This finding suggests an inherent limb-brightening that was not observable before due to limited angular resolution. Considering that the jet is viewed at an angle of $\sim50^\circ$, the observed limb-brightening is challenging to reconcile with the magnetohydrodynamic models and simulations, which predict that the Doppler-boosted jet edges should dominate over the non-boosted central layer. The conventional jet model that proposes a fast spine and a slow sheath with uniform transverse emissivity may pertain to our observations. However, in this model, the relativistic spine would need to travel at speeds of $\Gamma\gtrsim6.0-12.9$ along the de-projected jet distance of (2.3-10.8) $\times 10^3$ gravitational radii from the black hole. We propose an alternative scenario that suggests higher emissivity at the jet boundary layer, resulting from more efficient particle acceleration or mass loading onto the jet edges, and consider prospects for future observations with even higher angular resolution., Comment: 25 pages, 12 figures, 1 table, accepted for publication in the Astrophysical Journal Letters

We demonstrate that large language models (LLMs) exhibit consistent value orientations despite adopting diverse personas, revealing a persistent inertia in their responses that remains stable across the variety of roles they are prompted to assume. To systematically explore this phenomenon, we introduce the role-play-at-scale methodology, which involves prompting LLMs with randomized, diverse personas and analyzing the macroscopic trend of their responses. Unlike previous works that simply feed these questions to LLMs as if testing human subjects, our role-play-at-scale methodology diagnoses inherent tendencies in a systematic and scalable manner by: (1) prompting the model to act in different random personas and (2) asking the same question multiple times for each random persona. This approach reveals consistent patterns in LLM responses across diverse role-play scenarios, indicating deeply encoded inherent tendencies. Our findings contribute to the discourse on value alignment in foundation models and demonstrate the efficacy of role-play-at-scale as a diagnostic tool for uncovering encoded biases in LLMs.

We present Text-driven Object-Centric Style Transfer (TEXTOC), a novel method that guides style transfer at an object-centric level using textual inputs. The core of TEXTOC is our Patch-wise Co-Directional (PCD) loss, meticulously designed for precise object-centric transformations that are closely aligned with the input text. This loss combines a patch directional loss for text-guided style direction and a patch distribution consistency loss for even CLIP embedding distribution across object regions. It ensures a seamless and harmonious style transfer across object regions. Key to our method are the Text-Matched Patch Selection (TMPS) and Pre-fixed Region Selection (PRS) modules for identifying object locations via text, eliminating the need for segmentation masks. Lastly, we introduce an Adaptive Background Preservation (ABP) loss to maintain the original style and structural essence of the image's background. This loss is applied to dynamically identified background areas. Extensive experiments underline the effectiveness of our approach in creating visually coherent and textually aligned style transfers., Comment: 18 pages, 16 figures

Understanding traffic behavior is crucial for enhancing the stable functioning and safety of transportation systems. Previous percolation-based transportation studies have analyzed transition behaviors from free-flow to traffic-jam states, with a focus on robustness and resilience during congestion. However, relatively less attention is paid to the percolation analysis of the free-flow states, specifically how free-flow clusters form and grow. In this study, we investigate the percolation patterns of two opposing traffic scenarios -- traffic jam state and free-flow state -- within the same road network using Chengdu taxi data and compare their percolating behaviors. Our analysis reveals differences between the two scenarios in the growth patterns of the giant connected component (GCC), which is captured by a persistent gap between the GCC size curves, particularly during peak hours. We attribute these disparities to a long-range spatial correlation of traffic speed within a road network. Empirically, we find distinct long-range spatial correlations in traffic, using rescaled taxi speeds on roads, and we examine their relationship with each percolation pattern. Our analysis provides an integrated view of traffic dynamics and uncovers intrinsic traffic correlations within urban areas that drive these intriguing percolation patterns. Our findings also offer valuable metrics for effective traffic management and accident prevention strategies, aligning with urban transportation safety and reliability goals. These insights are beneficial for assessing and designing resilient urban road networks that maintain functionality under stress, ultimately improving the reliability of traffic assessments and reducing accidents., Comment: 17 pages, 10 figures

This paper introduces PeriodWave-Turbo, a high-fidelity and high-efficient waveform generation model via adversarial flow matching optimization. Recently, conditional flow matching (CFM) generative models have been successfully adopted for waveform generation tasks, leveraging a single vector field estimation objective for training. Although these models can generate high-fidelity waveform signals, they require significantly more ODE steps compared to GAN-based models, which only need a single generation step. Additionally, the generated samples often lack high-frequency information due to noisy vector field estimation, which fails to ensure high-frequency reproduction. To address this limitation, we enhance pre-trained CFM-based generative models by incorporating a fixed-step generator modification. We utilized reconstruction losses and adversarial feedback to accelerate high-fidelity waveform generation. Through adversarial flow matching optimization, it only requires 1,000 steps of fine-tuning to achieve state-of-the-art performance across various objective metrics. Moreover, we significantly reduce inference speed from 16 steps to 2 or 4 steps. Additionally, by scaling up the backbone of PeriodWave from 29M to 70M parameters for improved generalization, PeriodWave-Turbo achieves unprecedented performance, with a perceptual evaluation of speech quality (PESQ) score of 4.454 on the LibriTTS dataset. Audio samples, source code and checkpoints will be available at https://github.com/sh-lee-prml/PeriodWave., Comment: 9 pages, 9 tables, 1 figure

Unlike traditional media, online news platforms allow users to consume content that suits their tastes and to facilitate interactions with other people. However, as more personalized consumption of information and interaction with like-minded users increase, ideological bias can inadvertently increase and contribute to the formation of echo chambers, reinforcing the polarization of opinions. Although the structural characteristics of polarization among different ideological groups in online spaces have been extensively studied, research into how these groups emotionally interact with each other has not been as thoroughly explored. From this perspective, we investigate both structural and affective polarization between news media user groups on Naver News, South Korea's largest online news portal, during the period of 2022 Korean presidential election. By utilizing the dataset comprising 333,014 articles and over 36 million user comments, we uncover two distinct groups of users characterized by opposing political leanings and reveal significant bias and polarization among them. Additionally, we reveal the existence of echo chambers within co-commenting networks and investigate the asymmetric affective interaction patterns between the two polarized groups. Classification task of news media articles based on the distinct comment response patterns support the notion that different political groups may employ distinct communication strategies. Our approach based on network analysis on large-scale comment dataset offers novel insights into characteristics of user polarization in the online news platforms and the nuanced interaction nature between user groups., Comment: 21 pages, 6 figures

We propose a new reachability learning framework for high-dimensional nonlinear systems, focusing on reach-avoid problems. These problems require computing the reach-avoid set, which ensures that all its elements can safely reach a target set despite any disturbance within pre-specified bounds. Our framework has two main parts: offline learning of a newly designed reach-avoid value function and post-learning certification. Compared to prior works, our new value function is Lipschitz continuous and its associated Bellman operator is a contraction mapping, both of which improve the learning performance. To ensure deterministic guarantees of our learned reach-avoid set, we introduce two efficient post-learning certification methods. Both methods can be used online for real-time local certification or offline for comprehensive certification. We validate our framework in a 12-dimensional crazyflie drone racing hardware experiment and a simulated 10-dimensional highway takeover example., Comment: Submitted, under review

Recently, universal waveform generation tasks have been investigated conditioned on various out-of-distribution scenarios. Although GAN-based methods have shown their strength in fast waveform generation, they are vulnerable to train-inference mismatch scenarios such as two-stage text-to-speech. Meanwhile, diffusion-based models have shown their powerful generative performance in other domains; however, they stay out of the limelight due to slow inference speed in waveform generation tasks. Above all, there is no generator architecture that can explicitly disentangle the natural periodic features of high-resolution waveform signals. In this paper, we propose PeriodWave, a novel universal waveform generation model. First, we introduce a period-aware flow matching estimator that can capture the periodic features of the waveform signal when estimating the vector fields. Additionally, we utilize a multi-period estimator that avoids overlaps to capture different periodic features of waveform signals. Although increasing the number of periods can improve the performance significantly, this requires more computational costs. To reduce this issue, we also propose a single period-conditional universal estimator that can feed-forward parallel by period-wise batch inference. Additionally, we utilize discrete wavelet transform to losslessly disentangle the frequency information of waveform signals for high-frequency modeling, and introduce FreeU to reduce the high-frequency noise for waveform generation. The experimental results demonstrated that our model outperforms the previous models both in Mel-spectrogram reconstruction and text-to-speech tasks. All source code will be available at \url{https://github.com/sh-lee-prml/PeriodWave}., Comment: 24 pages, 16 tables, 4 figures

The explosive arrival of OpenAI's ChatGPT has fueled the globalization of large language model (LLM), which consists of billions of pretrained parameters that embodies the aspects of syntax and semantics. HyperAccel introduces latency processing unit (LPU), a latency-optimized and highly scalable processor architecture for the acceleration of LLM inference. LPU perfectly balances the memory bandwidth and compute logic with streamlined dataflow to maximize performance and efficiency. LPU is equipped with expandable synchronization link (ESL) that hides data synchronization latency between multiple LPUs. HyperDex complements LPU as an intuitive software framework to run LLM applications. LPU achieves 1.25 ms/token and 20.9 ms/token for 1.3B and 66B model, respectively, which is 2.09x and 1.37x faster than the GPU. LPU, synthesized using Samsung 4nm process, has total area of 0.824 mm2 and power consumption of 284.31 mW. LPU-based servers achieve 1.33x and 1.32x energy efficiency over NVIDIA H100 and L4 servers, respectively.

Complex urban road networks with high vehicle occupancy frequently face severe traffic congestion. Designing an effective strategy for managing multiple traffic lights plays a crucial role in managing congestion. However, most current traffic light management systems rely on human-crafted decisions, which may not adapt well to diverse traffic patterns. In this paper, we delve into two pivotal design components of the traffic light management system that can be dynamically adjusted to various traffic conditions: phase combination and phase time allocation. While numerous studies have sought an efficient strategy for managing traffic lights, most of these approaches consider a fixed traffic pattern and are limited to relatively small road networks. To overcome these limitations, we introduce a novel and practical framework to formulate the optimization of such design components using an offline meta black-box optimization. We then present a simple yet effective method to efficiently find a solution for the aforementioned problem. In our framework, we first collect an offline meta dataset consisting of pairs of design choices and corresponding congestion measures from various traffic patterns. After collecting the dataset, we employ the Attentive Neural Process (ANP) to predict the impact of the proposed design on congestion across various traffic patterns with well-calibrated uncertainty. Finally, Bayesian optimization, with ANP as a surrogate model, is utilized to find an optimal design for unseen traffic patterns through limited online simulations. Our experiment results show that our method outperforms state-of-the-art baselines on complex road networks in terms of the number of waiting vehicles. Surprisingly, the deployment of our method into a real-world traffic system was able to improve traffic throughput by 4.80\% compared to the original strategy., Comment: 12 pages, 7 figures, 10 tables

Time series generation is widely used in real-world applications such as simulation, data augmentation, and hypothesis test techniques. Recently, diffusion models have emerged as the de facto approach for time series generation, emphasizing diverse synthesis scenarios based on historical or correlated time series data streams. Since time series have unique characteristics, such as fixed time order and data scaling, standard Gaussian prior might be ill-suited for general time series generation. In this paper, we exploit the usage of diverse prior distributions for synthesis. Then, we propose TimeBridge, a framework that enables flexible synthesis by leveraging diffusion bridges to learn the transport between chosen prior and data distributions. Our model covers a wide range of scenarios in time series diffusion models, which leverages (i) data- and time-dependent priors for unconditional synthesis, and (ii) data-scale preserving synthesis with a constraint as a prior for conditional generation. Experimentally, our model achieves state-of-the-art performance in both unconditional and conditional time series generation tasks.

Quiescent galaxies are predominantly observed in local galaxy clusters. However, the fraction of quiescent galaxies in high-redshift clusters significantly varies among different clusters. In this study, we present the results of an analysis of the star formation (SF) properties of $z \sim 0.87$ clusters and groups from a cosmological hydrodynamical simulation Horizon Run 5. We investigate the correlation between the quiescent galaxy fraction (QF) of these model clusters/groups and their various internal or external properties. We find that halo mass is one of the most important characteristics as higher mass clusters and groups have higher QFs. We also find that other properties such as stellar-mass ratio and Friends-of-Friends fraction, which measures the proportion of the area around a cluster occupied by dense structures, may mildly affect the QFs of clusters and groups. This may indicate that the evolutionary history as well as the large-scale environment of clusters and groups also play a certain role in determining the SF status of high-redshift galaxy clusters and groups., Comment: 8 pages, 9 figures

Diffusion models have shown remarkable performance in generation problems over various domains including images, videos, text, and audio. A practical bottleneck of diffusion models is their sampling speed, due to the repeated evaluation of score estimation networks during the inference. In this work, we propose a novel framework capable of adaptively allocating compute required for the score estimation, thereby reducing the overall sampling time of diffusion models. We observe that the amount of computation required for the score estimation may vary along the time step for which the score is estimated. Based on this observation, we propose an early-exiting scheme, where we skip the subset of parameters in the score estimation network during the inference, based on a time-dependent exit schedule. Using the diffusion models for image synthesis, we show that our method could significantly improve the sampling throughput of the diffusion models without compromising image quality. Furthermore, we also demonstrate that our method seamlessly integrates with various types of solvers for faster sampling, capitalizing on their compatibility to enhance overall efficiency. The source code and our experiments are available at \url{https://github.com/taehong-moon/ee-diffusion}, Comment: ICML 2024

Research on hate speech has predominantly revolved around detection and interpretation from textual inputs, leaving verbal content largely unexplored. While there has been limited exploration into hate speech detection within verbal acoustic speech inputs, the aspect of interpretability has been overlooked. Therefore, we introduce a new task of explainable audio hate speech detection. Specifically, we aim to identify the precise time intervals, referred to as audio frame-level rationales, which serve as evidence for hate speech classification. Towards this end, we propose two different approaches: cascading and End-to-End (E2E). The cascading approach initially converts audio to transcripts, identifies hate speech within these transcripts, and subsequently locates the corresponding audio time frames. Conversely, the E2E approach processes audio utterances directly, which allows it to pinpoint hate speech within specific time frames. Additionally, due to the lack of explainable audio hate speech datasets that include audio frame-level rationales, we curated a synthetic audio dataset to train our models. We further validated these models on actual human speech utterances and found that the E2E approach outperforms the cascading method in terms of the audio frame Intersection over Union (IoU) metric. Furthermore, we observed that including frame-level rationales significantly enhances hate speech detection accuracy for the E2E approach. \textbf{Disclaimer} The reader may encounter content of an offensive or hateful nature. However, given the nature of the work, this cannot be avoided., Comment: Accepted to SIGDIAL 2024

Recent dialogue systems rely on turn-based spoken interactions, requiring accurate Automatic Speech Recognition (ASR). Errors in ASR can significantly impact downstream dialogue tasks. To address this, using dialogue context from user and agent interactions for transcribing subsequent utterances has been proposed. This method incorporates the transcription of the user's speech and the agent's response as model input, using the accumulated context generated by each turn. However, this context is susceptible to ASR errors because it is generated by the ASR model in an auto-regressive fashion. Such noisy context can further degrade the benefits of context input, resulting in suboptimal ASR performance. In this paper, we introduce Context Noise Representation Learning (CNRL) to enhance robustness against noisy context, ultimately improving dialogue speech recognition accuracy. To maximize the advantage of context awareness, our approach includes decoder pre-training using text-based dialogue data and noise representation learning for a context encoder. Based on the evaluation of speech dialogues, our method shows superior results compared to baselines. Furthermore, the strength of our approach is highlighted in noisy environments where user speech is barely audible due to real-world noise, relying on contextual information to transcribe the input accurately., Comment: 11 pages, 2 figures, Accepted to SIGDIAL2024

Accurate patient selection is critical in radiotherapy (RT) to prevent ineffective treatments. Traditional survival prediction models, relying on structured data, often lack precision. This study explores the potential of large language models (LLMs) to structure unstructured electronic health record (EHR) data, thereby improving survival prediction accuracy through comprehensive clinical information integration. Data from 34,276 patients treated with RT at Yonsei Cancer Center between 2013 and 2023 were analyzed, encompassing both structured and unstructured data. An open-source LLM was used to structure the unstructured EHR data via single-shot learning, with its performance compared against a domain-specific medical LLM and a smaller variant. Survival prediction models were developed using statistical, machine learning, and deep learning approaches, incorporating both structured and LLM-structured data. Clinical experts evaluated the accuracy of the LLM-structured data. The open-source LLM achieved 87.5% accuracy in structuring unstructured EHR data without additional training, significantly outperforming the domain-specific medical LLM, which reached only 35.8% accuracy. Larger LLMs were more effective, particularly in extracting clinically relevant features like general condition and disease extent, which closely correlated with patient survival. Incorporating LLM-structured clinical features into survival prediction models significantly improved accuracy, with the C-index of deep learning models increasing from 0.737 to 0.820. These models also became more interpretable by emphasizing clinically significant factors. This study shows that general-domain LLMs, even without specific medical training, can effectively structure large-scale unstructured EHR data, substantially enhancing the accuracy and interpretability of clinical predictive models., Comment: 23 pages, 2 tables, 4 figures

Using 1533 type Ia supernovae (SNe Ia) from the five-year sample of the Dark Energy Survey (DES), we investigate the effects of projected galactocentric separation between the SNe and their host galaxies on their light curves and standardization. We show, for the first time, that the difference in SN Ia post-standardization brightnesses between high and low-mass hosts reduces from $0.078\pm0.011$ mag in the full sample to $0.036 \pm 0.018$ mag for SNe Ia located in the outer regions of their host galaxies, while increasing to $0.100 \pm 0.014$ mag for SNe in the inner regions. In these inner regions, the step can be reduced (but not removed) using a model where the $R_V$ of dust along the line-of-sight to the SN changes as a function of galaxy properties. To explain the remaining difference, we use the distributions of the SN Ia stretch parameter to test whether the inferred age of SN progenitors are more varied in the inner regions of galaxies. We find that the proportion of high-stretch SNe Ia in red (older) environments is more prominent in outer regions and that the outer regions stretch distributions are overall more homogeneous compared to inner regions, but conclude that this effect cannot explain the reduction in significance of any Hubble residual step in outer regions. We conclude that the standardized distances of SNe Ia located in the outer regions of galaxies are less affected by their global host galaxy properties than those in the inner regions., Comment: 17 pages, 13 figures

In this paper, we introduce a memory-efficient CNN (convolutional neural network), which enables resource-constrained low-end embedded and IoT devices to perform on-device vision tasks, such as image classification and object detection, using extremely low memory, i.e., only 63 KB on ImageNet classification. Based on the bottleneck block of MobileNet, we propose three design principles that significantly curtail the peak memory usage of a CNN so that it can fit the limited KB memory of the low-end device. First, 'input segmentation' divides an input image into a set of patches, including the central patch overlapped with the others, reducing the size (and memory requirement) of a large input image. Second, 'patch tunneling' builds independent tunnel-like paths consisting of multiple bottleneck blocks per patch, penetrating through the entire model from an input patch to the last layer of the network, maintaining lightweight memory usage throughout the whole network. Lastly, 'bottleneck reordering' rearranges the execution order of convolution operations inside the bottleneck block such that the memory usage remains constant regardless of the size of the convolution output channels. The experiment result shows that the proposed network classifies ImageNet with extremely low memory (i.e., 63 KB) while achieving competitive top-1 accuracy (i.e., 61.58\%). To the best of our knowledge, the memory usage of the proposed network is far smaller than state-of-the-art memory-efficient networks, i.e., up to 89x and 3.1x smaller than MobileNet (i.e., 5.6 MB) and MCUNet (i.e., 196 KB), respectively.

We introduce EXAONE 3.0 instruction-tuned language model, the first open model in the family of Large Language Models (LLMs) developed by LG AI Research. Among different model sizes, we publicly release the 7.8B instruction-tuned model to promote open research and innovations. Through extensive evaluations across a wide range of public and in-house benchmarks, EXAONE 3.0 demonstrates highly competitive real-world performance with instruction-following capability against other state-of-the-art open models of similar size. Our comparative analysis shows that EXAONE 3.0 excels particularly in Korean, while achieving compelling performance across general tasks and complex reasoning. With its strong real-world effectiveness and bilingual proficiency, we hope that EXAONE keeps contributing to advancements in Expert AI. Our EXAONE 3.0 instruction-tuned model is available at https://huggingface.co/LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct

We present GRAFX, an open-source library designed for handling audio processing graphs in PyTorch. Along with various library functionalities, we describe technical details on the efficient parallel computation of input graphs, signals, and processor parameters in GPU. Then, we show its example use under a music mixing scenario, where parameters of every differentiable processor in a large graph are optimized via gradient descent. The code is available at https://github.com/sh-lee97/grafx., Comment: Accepted to DAFx 2024 demo

Recent advancements in low-cost ensemble learning have demonstrated improved efficiency for image classification. However, the existing low-cost ensemble methods show relatively lower accuracy compared to conventional ensemble learning. In this paper, we propose a new low-cost ensemble learning, which can simultaneously achieve high efficiency and classification performance. A CNN is transformed into a multi-branch structure without introduction of additional components, which maintains the computational complexity as that of the original single model and also enhances diversity among the branches' outputs via sufficient separation between different pathways of the branches. In addition, we propose a new strategy that applies grouped convolution in the branches with different numbers of groups in different branches, which boosts the diversity of the branches' outputs. For training, we employ knowledge distillation using the ensemble of the outputs as the teacher signal. The high diversity among the outputs enables to form a powerful teacher, enhancing the individual branch's classification performance and consequently the overall ensemble performance. Experimental results show that our method achieves state-of-the-art classification accuracy and higher uncertainty estimation performance compared to previous low-cost ensemble methods. The code is available at https://github.com/hjdw2/SEMBG.

Recent ensemble learning methods for image classification have been shown to improve classification accuracy with low extra cost. However, they still require multiple trained models for ensemble inference, which eventually becomes a significant burden when the model size increases. In this paper, we propose a low-cost ensemble learning and inference, called Network Fission Ensembles (NFE), by converting a conventional network itself into a multi-exit structure. Starting from a given initial network, we first prune some of the weights to reduce the training burden. We then group the remaining weights into several sets and create multiple auxiliary paths using each set to construct multi-exits. We call this process Network Fission. Through this, multiple outputs can be obtained from a single network, which enables ensemble learning. Since this process simply changes the existing network structure to multi-exits without using additional networks, there is no extra computational burden for ensemble learning and inference. Moreover, by learning from multiple losses of all exits, the multi-exits improve performance via regularization, and high performance can be achieved even with increased network sparsity. With our simple yet effective method, we achieve significant improvement compared to existing ensemble methods. The code is available at https://github.com/hjdw2/NFE.

Floquet engineering of topological phase transitions driven by a high-frequency time-periodic field is a promising approach to realizing new topological phases of matter distinct from static states. Here, we theoretically investigate Floquet engineering topological phase transitions in the quantum spin Hall $\alpha$-$T_{3}$ system driven by an off-resonant circularly polarized light. In addition to the quantum spin (anomalous) Hall insulator phase with multiple helical (chiral) edge states, spin-polarized topological metallic phases are expected, where the bulk topological band gap of one spin sub-band overlaps with the other gapless spin sub-band. Moreover, with a staggered potential, the topological invariants of the system depend on whether the middle band is occupied because of the breaking of particle-hole symmetry. Our work highlights the significance of Floquet engineering in realizing new topological phases in the $\alpha$-$T_{3}$ lattice., Comment: 10 pages, 5 figures

Spin- and charge-lattice interactions are potential key factors in the microscopic mechanism of high-temperature superconductivity in cuprates. Although both interactions can dramatically shape the low-energy electronic structure, their phenomenological roles in superconductivity are usually investigated independently. Employing angle-resolved photoemission spectroscopy, we reveal the spectroscopic fingerprint of short-range antiferromagnetic order in conjunction with enhanced electron-phonon interaction in the electron-doped cuprate superconductor $\mathrm{Nd_{1.85}Ce_{0.15}CuO_4}$. The observed mode coupling exhibits a strong momentum dependence that is in striking contrast to the node-antinode dichotomy previously observed in the hole-doped cuprates. Our results reveal an intimate relationship between electron-phonon coupling and antiferromagnetic fluctuations, which collectively sets the stage for unconventional superconductivity in the electron-doped cuprates.

BaCo$_2$(AsO$_4$)$_2$ (BCAO), a honeycomb cobaltate, is considered a promising candidate for materials displaying the Kitaev quantum spin liquid state. This assumption is based on the distinctive characteristics of Co$^{2+}$ ions (3$d^7$) within an octahedral crystal environment, resulting in spin-orbit-coupled $J_{\rm eff}$~=~1/2 doublet states. However, recent experimental observations and theoretical analyses have raised questions regarding this hypothesis. Despite these uncertainties, reports of continuum excitations reminiscent of spinon excitations have prompted further investigations. In this study, we explore the magnetic phases of BCAO under both in-plane and out-of-plane magnetic fields, employing dc and ac magnetic susceptibilities, capacitance, and torque magnetometry measurement. Our results affirm the existence of multiple field-induced magnetic phases, with strong anisotropy of the phase boundaries between in-plane and out-of-plane fields. To elucidate the nature of these phases, we develop a minimal anisotropic exchange model. This model, supported by combined first principles calculations and theoretical modeling, quantitatively reproduces our experimental data. In BCAO, the combination of strong bond-independent XXZ anisotropy and geometric frustration leads to significant quantum order by disorder effects that stabilize collinear phases under both zero and finite magnetic fields., Comment: 14 pages, 5 main + 5 supplementary figures