Search

We acquired 450 {\mu}m and 850 {\mu}m dust continuum polarization observations toward the inner region of the Central Molecular Zone (CMZ) as part of the B-Fields In Star-Forming Region Observations (BISTRO) survey using the POL-2 polarimeter on the James Clerk Maxwell Telescope. These observations encompassed three dense structures: the 20 km s{^{-1}} cloud (20MC), 50 km s{^{-1}} cloud (50MC), and circumnuclear disk (CND). Our aim is to investigate the magnetic field morphology and strength in the inner region of the CMZ using polarized dust continuum and the Davis-Chandrasekhar-Fermi method. The magnetic field morphology is highly ordered in all three dense regions. The plane-of-sky magnetic field strengths are {\sim}1 mG for the 20MC and the 50MC, and {\sim}2 mG for the CND. We compare the energy contributions of turbulence, gravity, and thermal motion with that of the magnetic field using the plasma {\beta}, mass-to-flux ratio, and Alfv\'en Mach number. The outcomes reveal the magnetic field stands out as the predominant factor within the inner region of the CMZ. The dominance of the magnetic field may explain the low star-forming rate in the CMZ. We further investigate the dust grain alignment efficiency by exploring the relationship between polarization fraction and total intensity. The results suggest that dust grains are well aligned with the magnetic fields., Comment: Accepted for publication in ApJ. 31 pages, 21 figures (20 in main text, 1 in appendix), 1 appendix

Model-based Reinforcement Learning (MBRL) has emerged as a promising paradigm for autonomous driving, where data efficiency and robustness are critical. Yet, existing solutions often rely on carefully crafted, task specific extrinsic rewards, limiting generalization to new tasks or environments. In this paper, we propose InDRiVE (Intrinsic Disagreement based Reinforcement for Vehicle Exploration), a method that leverages purely intrinsic, disagreement based rewards within a Dreamer based MBRL framework. By training an ensemble of world models, the agent actively explores high uncertainty regions of environments without any task specific feedback. This approach yields a task agnostic latent representation, allowing for rapid zero shot or few shot fine tuning on downstream driving tasks such as lane following and collision avoidance. Experimental results in both seen and unseen environments demonstrate that InDRiVE achieves higher success rates and fewer infractions compared to DreamerV2 and DreamerV3 baselines despite using significantly fewer training steps. Our findings highlight the effectiveness of purely intrinsic exploration for learning robust vehicle control behaviors, paving the way for more scalable and adaptable autonomous driving systems., Comment: This work has been submitted to IROS 2025 and is currently under review

Graph processing requires irregular, fine-grained random access patterns incompatible with contemporary off-chip memory architecture, leading to inefficient data access. This inefficiency makes graph processing an extremely memory-bound application. Because of this, existing graph processing accelerators typically employ a graph tiling-based or processing-in-memory (PIM) approach to relieve the memory bottleneck. In the tiling-based approach, a graph is split into chunks that fit within the on-chip cache to maximize data reuse. In the PIM approach, arithmetic units are placed within memory to perform operations such as reduction or atomic addition. However, both approaches have several limitations, especially when implemented on current memory standards (i.e., DDR). Because the access granularity provided by DDR is much larger than that of the graph vertex property data, much of the bandwidth and cache capacity are wasted. PIM is meant to alleviate such issues, but it is difficult to use in conjunction with the tiling-based approach, resulting in a significant disadvantage. Furthermore, placing arithmetic units inside a memory chip is expensive, thereby supporting multiple types of operation is thought to be impractical. To address the above limitations, we present Piccolo, an end-to-end efficient graph processing accelerator with fine-grained in-memory random scatter-gather. Instead of placing expensive arithmetic units in off-chip memory, Piccolo focuses on reducing the off-chip traffic with non-arithmetic function-in-memory of random scatter-gather. To fully benefit from in-memory scatter-gather, Piccolo redesigns the cache and MHA of the accelerator such that it can enjoy both the advantage of tiling and in-memory operations. Piccolo achieves a maximum speedup of 3.28$\times$ and a geometric mean speedup of 1.62$\times$ across various and extensive benchmarks., Comment: HPCA 2025

Video anomaly detection (VAD) is crucial for video analysis and surveillance in computer vision. However, existing VAD models rely on learned normal patterns, which makes them difficult to apply to diverse environments. Consequently, users should retrain models or develop separate AI models for new environments, which requires expertise in machine learning, high-performance hardware, and extensive data collection, limiting the practical usability of VAD. To address these challenges, this study proposes customizable video anomaly detection (C-VAD) technique and the AnyAnomaly model. C-VAD considers user-defined text as an abnormal event and detects frames containing a specified event in a video. We effectively implemented AnyAnomaly using a context-aware visual question answering without fine-tuning the large vision language model. To validate the effectiveness of the proposed model, we constructed C-VAD datasets and demonstrated the superiority of AnyAnomaly. Furthermore, our approach showed competitive performance on VAD benchmark datasets, achieving state-of-the-art results on the UBnormal dataset and outperforming other methods in generalization across all datasets. Our code is available online at github.com/SkiddieAhn/Paper-AnyAnomaly.

We report a precise measurement of the ratio of branching fractions $\mathcal{B}(\Lambda_c^+\to p K_S^0 \pi^0)/\mathcal{B}(\Lambda_c^+\to p K^- \pi^+)$ using 980 fb$^{-1}$ of $e^+e^-$ data from the Belle experiment. We obtain a value of $\mathcal{B}(\Lambda_c^+\to p K_S^0 \pi^0)/\mathcal{B}(\Lambda_c^+\to p K^- \pi^+)=0.339\pm 0.002\pm 0.009$, where the first and second uncertainties are statistical and systematic, respectively. This Belle result is consistent with the previous measurement from the CLEO experiment but has a fivefold improvement in precision. By combining our result with the world average $\mathcal{B}(\Lambda_c^+\to p K^- \pi^+)$, we obtain the absolute branching fraction $\mathcal{B}(\Lambda_c^+\to p K_S^0 \pi^0)=(2.12\pm 0.01\pm 0.05 \pm 0.10)\%$, where the uncertainties are statistical, systematic, and the uncertainty in the absolute branching fraction scale $\mathcal{B}(\Lambda_c^+\to p K^- \pi^+)$, respectively. This measurement can shed light on hadronic decay mechanisms in charmed baryon decays., Comment: 20 pages, 7 figures

Limited data is a common problem in remote sensing due to the high cost of obtaining annotated samples. In the few-shot segmentation task, models are typically trained on base classes with abundant annotations and later adapted to novel classes with limited examples. However, this often necessitates specialized model architectures or complex training strategies. Instead, we propose a simple approach that leverages diffusion models to generate diverse variations of novel-class objects within a given scene, conditioned by the limited examples of the novel classes. By framing the problem as an image inpainting task, we synthesize plausible instances of novel classes under various environments, effectively increasing the number of samples for the novel classes and mitigating overfitting. The generated samples are then assessed using a cosine similarity metric to ensure semantic consistency with the novel classes. Additionally, we employ Segment Anything Model (SAM) to segment the generated samples and obtain precise annotations. By using high-quality synthetic data, we can directly fine-tune off-the-shelf segmentation models. Experimental results demonstrate that our method significantly enhances segmentation performance in low-data regimes, highlighting its potential for real-world remote sensing applications., Comment: Accepted to ICLRW 2025 (Oral)

This paper presents a novel Perceptual Motor Learning (PML) framework integrated with Active Inference (AIF) to enhance lateral control in Highly Automated Vehicles (HAVs). PML, inspired by human motor learning, emphasizes the seamless integration of perception and action, enabling efficient decision-making in dynamic environments. Traditional autonomous driving approaches--including modular pipelines, imitation learning, and reinforcement learning--struggle with adaptability, generalization, and computational efficiency. In contrast, PML with AIF leverages a generative model to minimize prediction error ("surprise") and actively shape vehicle control based on learned perceptual-motor representations. Our approach unifies deep learning with active inference principles, allowing HAVs to perform lane-keeping maneuvers with minimal data and without extensive retraining across different environments. Extensive experiments in the CARLA simulator demonstrate that PML with AIF enhances adaptability without increasing computational overhead while achieving performance comparable to conventional methods. These findings highlight the potential of PML-driven active inference as a robust alternative for real-world autonomous driving applications., Comment: This work has been submitted to IROS 2025 and is currently under review. arXiv admin note: text overlap with arXiv:2407.07684

Point cloud registration is important in computer-aided interventions (CAI). While learning-based point cloud registration methods have been developed, their clinical application is hampered by issues of generalizability and explainability. Therefore, classical point cloud registration methods, such as Iterative Closest Point (ICP), are still widely applied in CAI. ICP methods fail to consider that: (1) the points have well-defined semantic meaning, in that each point can be related to a specific anatomical label; (2) the deformation needs to follow biomechanical energy constraints. In this paper, we present a novel semantic ICP (sem-ICP) method that handles multiple point labels and uses linear elastic energy regularization. We use semantic labels to improve the robustness of the closest point matching and propose a new point cloud deformation representation to apply explicit biomechanical energy regularization. Our experiments on the Learn2reg abdominal MR-CT registration dataset and a trans-oral robotic surgery ultrasound-CT registration dataset show that our method improves the Hausdorff distance compared with other state-of-the-art ICP-based registration methods. We also perform a sensitivity study to show that our rigid initialization achieves better convergence with different initializations and visible ratios., Comment: 10 pages, 3 figures, submitted to MICCAI 2025

End-to-end vision-based imitation learning has demonstrated promising results in autonomous driving by learning control commands directly from expert demonstrations. However, traditional approaches rely on either regressionbased models, which provide precise control but lack confidence estimation, or classification-based models, which offer confidence scores but suffer from reduced precision due to discretization. This limitation makes it challenging to quantify the reliability of predicted actions and apply corrections when necessary. In this work, we introduce a dual-head neural network architecture that integrates both regression and classification heads to improve decision reliability in imitation learning. The regression head predicts continuous driving actions, while the classification head estimates confidence, enabling a correction mechanism that adjusts actions in low-confidence scenarios, enhancing driving stability. We evaluate our approach in a closed-loop setting within the CARLA simulator, demonstrating its ability to detect uncertain actions, estimate confidence, and apply real-time corrections. Experimental results show that our method reduces lane deviation and improves trajectory accuracy by up to 50%, outperforming conventional regression-only models. These findings highlight the potential of classification-guided confidence estimation in enhancing the robustness of vision-based imitation learning for autonomous driving. The source code is available at https://github.com/ElaheDlv/Confidence_Aware_IL.

Instance detection (InsDet) aims to localize specific object instances within a novel scene imagery based on given visual references. Technically, it requires proposal detection to identify all possible object instances, followed by instance-level matching to pinpoint the ones of interest. Its open-world nature supports its wide-ranging applications from robotics to AR/VR, but also presents significant challenges: methods must generalize to unknown testing data distributions because (1) the testing scene imagery is unseen during training, and (2) there are domain gaps between visual references and detected proposals. Existing methods attempt to tackle these challenges by synthesizing diverse training examples or utilizing off-the-shelf foundation models (FMs). However, they only partially capitalize the available open-world information. In this paper, we approach InsDet from an Open-World perspective, introducing our method IDOW. We find that, while pretrained FMs yield high recall in instance detection, they are not specifically optimized for instance-level feature matching. To address this, we adapt pretrained FMs for improved instance-level matching using open-world data. Our approach incorporates metric learning along with novel data augmentations, which sample distractors as negative examples and synthesize novel-view instances to enrich the visual references. Extensive experiments demonstrate that our method significantly outperforms prior works, achieving >10 AP over previous results on two recently released challenging benchmark datasets in both conventional and novel instance detection settings., Comment: Accepted at CVPR 2025

Recent advancements in instruction fine-tuning, alignment methods such as reinforcement learning from human feedback (RLHF), and optimization techniques like direct preference optimization (DPO) have significantly enhanced the adaptability of large language models (LLMs) to user preferences. However, despite these innovations, many LLMs continue to exhibit biases toward Western, Anglo-centric, or American cultures, with performance on English data consistently surpassing that of other languages. This reveals a persistent cultural gap in LLMs, which complicates their ability to accurately process culturally rich and diverse figurative language such as proverbs. To address this, we introduce Jawaher, a benchmark designed to assess LLMs' capacity to comprehend and interpret Arabic proverbs. Jawaher includes proverbs from various Arabic dialects, along with idiomatic translations and explanations. Through extensive evaluations of both open- and closed-source models, we find that while LLMs can generate idiomatically accurate translations, they struggle with producing culturally nuanced and contextually relevant explanations. These findings highlight the need for ongoing model refinement and dataset expansion to bridge the cultural gap in figurative language processing., Comment: Project GitHub page is accessible at: https://github.com/UBC-NLP/jawaher

Deep neural networks trained using gradient descent with a fixed learning rate $\eta$ often operate in the regime of "edge of stability" (EOS), where the largest eigenvalue of the Hessian equilibrates about the stability threshold $2/\eta$. In this work, we present a fine-grained analysis of the learning dynamics of (deep) linear networks (DLNs) within the deep matrix factorization loss beyond EOS. For DLNs, loss oscillations beyond EOS follow a period-doubling route to chaos. We theoretically analyze the regime of the 2-period orbit and show that the loss oscillations occur within a small subspace, with the dimension of the subspace precisely characterized by the learning rate. The crux of our analysis lies in showing that the symmetry-induced conservation law for gradient flow, defined as the balancing gap among the singular values across layers, breaks at EOS and decays monotonically to zero. Overall, our results contribute to explaining two key phenomena in deep networks: (i) shallow models and simple tasks do not always exhibit EOS; and (ii) oscillations occur within top features. We present experiments to support our theory, along with examples demonstrating how these phenomena occur in nonlinear networks and how they differ from those which have benign landscape such as in DLNs., Comment: Published in ICLR 2025

The inherent complexity of boundary plasma, characterized by multi-scale and multi-physics challenges, has historically restricted high-fidelity simulations to scientific research due to their intensive computational demands. Consequently, routine applications such as discharge control and scenario development have relied on faster, but less accurate empirical methods. This work introduces DivControlNN, a novel machine-learning-based surrogate model designed to address these limitations by enabling quasi-real-time predictions (i.e., $\sim0.2$ ms) of boundary and divertor plasma behavior. Trained on over 70,000 2D UEDGE simulations from KSTAR tokamak equilibria, DivControlNN employs latent space mapping to efficiently represent complex divertor plasma states, achieving a computational speed-up of over $10^8$ compared to traditional simulations while maintaining a relative error below 20% for key plasma property predictions. During the 2024 KSTAR experimental campaign, a prototype detachment control system powered by DivControlNN successfully demonstrated detachment control on its first attempt, even for a new tungsten divertor configuration and without any fine-tuning. These results highlight the transformative potential of DivControlNN in overcoming diagnostic challenges in future fusion reactors by providing fast, robust, and reliable predictions for advanced integrated control systems., Comment: 40 pages, 18 figures

Identifying AI-generated content is critical for the safe and ethical use of generative AI. Recent research has focused on developing detectors that generalize to unknown generators, with popular methods relying either on high-level features or low-level fingerprints. However, these methods have clear limitations: biased towards unseen content, or vulnerable to common image degradations, such as JPEG compression. To address these issues, we propose a novel approach, SFLD, which incorporates PatchShuffle to integrate high-level semantic and low-level textural information. SFLD applies PatchShuffle at multiple levels, improving robustness and generalization across various generative models. Additionally, current benchmarks face challenges such as low image quality, insufficient content preservation, and limited class diversity. In response, we introduce TwinSynths, a new benchmark generation methodology that constructs visually near-identical pairs of real and synthetic images to ensure high quality and content preservation. Our extensive experiments and analysis show that SFLD outperforms existing methods on detecting a wide variety of fake images sourced from GANs, diffusion models, and TwinSynths, demonstrating the state-of-the-art performance and generalization capabilities to novel generative models., Comment: IEEE/CVF WACV 2025, Oral

In recent neutrino detectors, neutrons produced in neutrino reactions play an important role. Muon capture on oxygen nuclei is one of the processes that produce neutrons in water Cherenkov detectors. We measured neutron multiplicity in the process using cosmic ray muons that stop in the gadolinium-loaded Super-Kamiokande detector. For this measurement, neutron detection efficiency is obtained with the muon capture events followed by gamma rays to be $50.2^{+2.0}_{-2.1}\%$. By fitting the observed multiplicity considering the detection efficiency, we measure neutron multiplicity in muon capture as $P(0)=24\pm3\%$, $P(1)=70^{+3}_{-2}\%$, $P(2)=6.1\pm0.5\%$, $P(3)=0.38\pm0.09\%$. This is the first measurement of the multiplicity of neutrons associated with muon capture without neutron energy threshold.

There are two main approaches to recent extractive summarization: the sentence-level framework, which selects sentences to include in a summary individually, and the summary-level framework, which generates multiple candidate summaries and ranks them. Previous work in both frameworks has primarily focused on improving which sentences in a document should be included in the summary. However, the sentence order of extractive summaries, which is critical for the quality of a summary, remains underexplored. In this paper, we introduce OrderSum, a novel extractive summarization model that semantically orders sentences within an extractive summary. OrderSum proposes a new representation method to incorporate the sentence order into the embedding of the extractive summary, and an objective function to train the model to identify which extractive summary has a better sentence order in the semantic space. Extensive experimental results demonstrate that OrderSum obtains state-of-the-art performance in both sentence inclusion and sentence order for extractive summarization. In particular, OrderSum achieves a ROUGE-L score of 30.52 on CNN/DailyMail, outperforming the previous state-of-the-art model by a large margin of 2.54., Comment: 29 pages, 12 pages for the main body, 5 figures

OpenNotes enables patients to access EHR notes, but medical jargon can hinder comprehension. To improve understanding, we evaluated closed- and open-source LLMs for extracting and prioritizing key medical terms using prompting, fine-tuning, and data augmentation. We assessed LLMs on 106 expert-annotated EHR notes, experimenting with (i) general vs. structured prompts, (ii) zero-shot vs. few-shot prompting, (iii) fine-tuning, and (iv) data augmentation. To enhance open-source models in low-resource settings, we used ChatGPT for data augmentation and applied ranking techniques. We incrementally increased the augmented dataset size (10 to 10,000) and conducted 5-fold cross-validation, reporting F1 score and Mean Reciprocal Rank (MRR). Our result show that fine-tuning and data augmentation improved performance over other strategies. GPT-4 Turbo achieved the highest F1 (0.433), while Mistral7B with data augmentation had the highest MRR (0.746). Open-source models, when fine-tuned or augmented, outperformed closed-source models. Notably, the best F1 and MRR scores did not always align. Few-shot prompting outperformed zero-shot in vanilla models, and structured prompts yielded different preferences across models. Fine-tuning improved zero-shot performance but sometimes degraded few-shot performance. Data augmentation performed comparably or better than other methods. Our evaluation highlights the effectiveness of prompting, fine-tuning, and data augmentation in improving model performance for medical jargon extraction in low-resource scenarios., Comment: 21pages, 5 figures, 4 tables

Penetration depth calculation quantifies the extent of overlap between two objects and is crucial in fields like simulations, the metaverse, and robotics. Recognizing its significance, efforts have been made to accelerate this computation using parallel computing resources, such as CPUs and GPUs. Unlike traditional GPU cores, modern GPUs incorporate specialized ray-tracing cores (RT-cores) primarily used for rendering applications. We introduce a novel algorithm for penetration depth calculation that leverages RT-cores. Our approach includes a ray-tracing based algorithm for penetration surface extraction and another for calculating Hausdorff distance, optimizing the use of RT-cores. We tested our method across various generations of RTX GPUs with different benchmark scenes. The results demonstrated that our algorithm outperformed a state-of-the-art penetration depth calculation method and conventional GPU implementations by up to 37.66 and 5.33 times, respectively. These findings demonstrate the efficiency of our RT core-based method and suggest broad applicability for RT-cores in diverse computational tasks., Comment: 10 pages, 8 figures, under review for a journal

Recent advancements in table-based reasoning have expanded beyond factoid-level QA to address insight-level tasks, where systems should synthesize implicit knowledge in the table to provide explainable analyses. Although effective, existing studies remain confined to scenarios where a single gold table is given alongside the user query, failing to address cases where users seek comprehensive insights from multiple unknown tables. To bridge these gaps, we propose MT-RAIG Bench, design to evaluate systems on Retrieval-Augmented Insight Generation over Mulitple-Tables. Additionally, to tackle the suboptimality of existing automatic evaluation methods in the table domain, we further introduce a fine-grained evaluation framework MT-RAIG Eval, which achieves better alignment with human quality judgments on the generated insights. We conduct extensive experiments and reveal that even frontier LLMs still struggle with complex multi-table reasoning, establishing our MT-RAIG Bench as a challenging testbed for future research., Comment: Work in progress

We present the magnetic field in the dense material of the Central Molecular Zone (CMZ) of the Milky Way, traced in 850 $\mu$m polarized dust emission as part of the James Clerk Maxwell Telescope (JCMT) B-fields In STar-forming Region Observations (BISTRO) Survey. We observe a highly ordered magnetic field across the CMZ between Sgr B2 and Sgr C, which is strongly preferentially aligned with the orbital gas flows within the clouds of the CMZ. We find that the observed relative orientations are non-random at a $>$99% confidence level and are consistent with models in which the magnetic field vectors are aligned within 30$^{o}$ to the gas flows in 3D. The deviations from aligned magnetic fields are most prominent at positive Galactic longitudes, where the CMZ clouds are more massive, denser, and more actively forming stars. Our observed strongly preferentially parallel magnetic field morphology leads us to hypothesize that in the absence of star formation, the magnetic field in the CMZ is entrained in the orbital gas flows around Sgr A$^{*}$, while gravitational collapse and feedback in star-forming regions can locally reorder the field. This magnetic field behavior is similar to that observed in the CMZ of the nuclear starburst galaxy NGC 253. This suggests that despite its current low star formation rate, the CMZ of the Milky Way is analogous to those of more distant, actively star-forming, galaxies., Comment: Accepted in ApJL. 16 pages, 9 figures (4 in main text, 5 in appendices), 5 appendices

We consider the off-policy selection and learning in contextual bandits where the learner aims to select or train a reward-maximizing policy using data collected by a fixed behavior policy. Our contribution is two-fold. First, we propose a novel off-policy selection method that leverages a new betting-based confidence bound applied to an inverse propensity weight sequence. Our theoretical analysis reveals that our method achieves a significantly better, variance-adaptive guarantee upon prior art. Second, we propose a novel and generic condition on the optimization objective for off-policy learning that strikes a difference balance in bias and variance. One special case that we call freezing tends to induce small variance, which is preferred in small-data regimes. Our analysis shows that they match the best existing guarantee. In our empirical study, our selection method outperforms existing methods, and freezing exhibits improved performance in small-sample regimes., Comment: 36 pages, 8 figures

For $\beta>1$, let $T_\beta:[0,1]\rightarrow [0,1)$ be the $\beta$-transformation. We consider an invariant $T_\beta$-orbit closure contained in a closed interval with diameter $1/\beta$, then define a function $\Xi(\alpha,\beta)$ by the supremum of such $T_\beta$-orbit with frequency $\alpha$ in base $\beta$, i.e., the maximum value in the $T_\beta$-orbit closure. This paper effectively determines the maximal domain of $\Xi$, and explicitly specifies all possible minimal intervals containing $T_\beta$-orbits. For Addendum: The paper mentioned in the title is completed by this Addendum., Comment: 17+2 pages, reference updated, Addendum added

Using data samples of 102 million Upsilon(1S) events and 158 million Upsilon(2S) events collected by the Belle detector at the KEKB asymmetric-energy $e^+e^-$ collider, we search for [udsccbar] pentaquark states decaying to Jpsi Lambda. Using the first observations of Upsilon(1S, 2S) inclusive decays to Jpsi Lambda, we find evidence of the P_ccbars(4459)0 state with a significance of 3.3 standard deviations, including statistical and systematic uncertainties. We measure the mass and width of the Pccbars(4459)0 to be (4471.7 +- 4.8 +- 0.6) MeV/c2 and (21.9 +- 13.1 +- 2.7) MeV, respectively. The branching fractions for P_ccbars(4459)0 production are measured to be B[Upsilon(1S) -> P_ccbars(4459)0/ Pbar_ccbars(4459)0 + anything] = (3.5 +- 2.0 +- 0.2)*10-6 and B[Upsilin(2S) -> P_ccbars(4459)0/ Pbar_ccbars(4459)0 +anything] = (2.9 +- 1.7 +- 0.4)*10-6. The inclusive branching fractions of Upsilon(1S, 2S) -> Jpsi Lambda/Lambdabar are measured to be B[Upsilin(1S) -> Jpsi Lambda/Lambdabar + anything] = (36.9 +- 5.3 +- 2.4)*10-6 and B[Upsilon(2S) -> Jpsi Lambda/Lambdabar + anything] = (22.3 +- 5.7 +- 3.1)*10-6. We measure the visible cross section $\sigma(e^+e^- \to J/psi \Lambda/\bar\Lambda$ + anything) = (90 +- 14 +- 6) fb for the continuum production at $\sqrt{s} = 10.52$ GeV. In all cases, the first uncertainties are statistical and the second are systematic., Comment: 8 pages, 3 figures

Depth completion, predicting dense depth maps from sparse depth measurements, is an ill-posed problem requiring prior knowledge. Recent methods adopt learning-based approaches to implicitly capture priors, but the priors primarily fit in-domain data and do not generalize well to out-of-domain scenarios. To address this, we propose a zero-shot depth completion method composed of an affine-invariant depth diffusion model and test-time alignment. We use pre-trained depth diffusion models as depth prior knowledge, which implicitly understand how to fill in depth for scenes. Our approach aligns the affine-invariant depth prior with metric-scale sparse measurements, enforcing them as hard constraints via an optimization loop at test-time. Our zero-shot depth completion method demonstrates generalization across various domain datasets, achieving up to a 21\% average performance improvement over the previous state-of-the-art methods while enhancing spatial understanding by sharpening scene details. We demonstrate that aligning a monocular affine-invariant depth prior with sparse metric measurements is a proven strategy to achieve domain-generalizable depth completion without relying on extensive training data. Project page: https://hyoseok1223.github.io/zero-shot-depth-completion/., Comment: AAAI 2025, Project page: https://hyoseok1223.github.io/zero-shot-depth-completion/

We introduce a generalizable and unified framework to synthesize view-consistent and temporally coherent avatars from a single image, addressing the challenging problem of single-image avatar generation. While recent methods employ diffusion models conditioned on human templates like depth or normal maps, they often struggle to preserve appearance information due to the discrepancy between sparse driving signals and the actual human subject, resulting in multi-view and temporal inconsistencies. Our approach bridges this gap by combining the reconstruction power of regression-based 3D human reconstruction with the generative capabilities of a diffusion model. The dense driving signal from the initial reconstructed human provides comprehensive conditioning, ensuring high-quality synthesis faithful to the reference appearance and structure. Additionally, we propose a unified framework that enables the generalization learned from novel pose synthesis on in-the-wild videos to naturally transfer to novel view synthesis. Our video-based diffusion model enhances disentangled synthesis with high-quality view-consistent renderings for novel views and realistic non-rigid deformations in novel pose animation. Results demonstrate the superior generalization ability of our method across in-domain and out-of-domain in-the-wild datasets. Project page: https://humansensinglab.github.io/GAS/

Effective search methods are crucial for improving the performance of deep generative models at test time. In this paper, we introduce a novel test-time search method, Neural Genetic Search (NGS), which incorporates the evolutionary mechanism of genetic algorithms into the generation procedure of deep models. The core idea behind NGS is its crossover, which is defined as parent-conditioned generation using trained generative models. This approach offers a versatile and easy-to-implement search algorithm for deep generative models. We demonstrate the effectiveness and flexibility of NGS through experiments across three distinct domains: routing problems, adversarial prompt generation for language models, and molecular design., Comment: 19 pages

Accelerating inference in Large Language Models (LLMs) is critical for real-time interactions, as they have been widely incorporated into real-world services. Speculative decoding, a fully algorithmic solution, has gained attention for improving inference speed by drafting and verifying tokens, thereby generating multiple tokens in a single forward pass. However, current drafting strategies usually require significant fine-tuning or have inconsistent performance across tasks. To address these challenges, we propose Hierarchy Drafting (HD), a novel lossless drafting approach that organizes various token sources into multiple databases in a hierarchical framework based on temporal locality. In the drafting step, HD sequentially accesses multiple databases to obtain draft tokens from the highest to the lowest locality, ensuring consistent acceleration across diverse tasks and minimizing drafting latency. Our experiments on Spec-Bench using LLMs with 7B and 13B parameters demonstrate that HD outperforms existing database drafting methods, achieving robust inference speedups across model sizes, tasks, and temperatures., Comment: Findings of NAACL 2025

Two-stage stochastic programs (2SPs) are important tools for making decisions under uncertainty. Decision-makers use contextual information to generate a set of scenarios to represent the true conditional distribution. However, the number of scenarios required is a barrier to implementing 2SPs, motivating the problem of generating a small set of surrogate scenarios that yield high-quality decisions when they represent uncertainty. Current scenario generation approaches do not leverage contextual information or do not address computational concerns. In response, we propose contextual scenario generation (CSG) to learn a mapping between the context and a set of surrogate scenarios of user-specified size. First, we propose a distributional approach that learns the mapping by minimizing a distributional distance between the predicted surrogate scenarios and the true contextual distribution. Second, we propose a task-based approach that aims to produce surrogate scenarios that yield high-quality decisions. The task-based approach uses neural architectures to approximate the downstream objective and leverages the approximation to search for the mapping. The proposed approaches apply to various problem structures and loosely only require efficient solving of the associated subproblems and 2SPs defined on the reduced scenario sets. Numerical experiments demonstrating the effectiveness of the proposed methods are presented., Comment: 47 pages, 10 figures

We present a measurement of the branching fraction of $B^+\to\tau^+\nu_\tau$ decays using $(387\pm6)\times 10^6$ $\Upsilon(4S)$ collected between 2019 and 2022 with the Belle II detector at the SuperKEKB $e^+e^-$ collider. We reconstruct the accompanying $B^-$ meson using the hadronic tagging method, while $B^+\to\tau^+\nu_\tau$ candidates are identified in the recoil. We find evidence for $B^+\to\tau^+\nu_\tau$ decays at 3.0 standard deviations, including systematic uncertainties. The measured branching fraction is $\mathcal{B}(B^+\to\tau^+\nu_\tau) = [1.24 \pm 0.41 (\text{stat.}) \pm 0.19 (\text{syst.})] \times 10^{-4}$.

We address the convergence problem in learning the Optimal Transport (OT) map, where the OT Map refers to a map from one distribution to another while minimizing the transport cost. Semi-dual Neural OT, a widely used approach for learning OT Maps with neural networks, often generates fake solutions that fail to transfer one distribution to another accurately. We identify a sufficient condition under which the max-min solution of Semi-dual Neural OT recovers the true OT Map. Moreover, to address cases when this sufficient condition is not satisfied, we propose a novel method, OTP, which learns both the OT Map and the Optimal Transport Plan, representing the optimal coupling between two distributions. Under sharp assumptions on the distributions, we prove that our model eliminates the fake solution issue and correctly solves the OT problem. Our experiments show that the OTP model recovers the optimal transport map where existing methods fail and outperforms current OT-based models in image-to-image translation tasks. Notably, the OTP model can learn stochastic transport maps when deterministic OT Maps do not exist, such as one-to-many tasks like colorization., Comment: 18 pages, 10 figures

We present the Dark Energy Spectroscopic Instrument (DESI) Strong Lens Foundry. We discovered $\sim 3500$ new strong gravitational lens candidates in the DESI Legacy Imaging Surveys using residual neural networks (ResNet). We observed a subset (51) of our candidates using the Hubble Space Telescope (HST). All of them were confirmed to be strong lenses. We also briefly describe spectroscopic follow-up observations by DESI and Keck NIRES programs. From this very rich dataset, a number of studies will be carried out, including evaluating the quality of the ResNet search candidates and lens modeling. In this paper, we present our initial effort in these directions. In particular, as a demonstration, we present the lens model for DESI-165.4754-06.0423, with imaging data from HST, and lens and source redshifts from DESI and Keck NIRES, respectively. In this effort, we have applied a \emph{fully} forward-modeling Bayesian approach (GIGA-Lens), using \emph{multiple} GPUs, for the first time in both regards, to a strong lens with HST data, or any high resolution imaging., Comment: 32 pages, 9 figures, 5 tables. Submitted to The Astrophysical Journal

Recent 4D dynamic scene editing methods require editing thousands of 2D images used for dynamic scene synthesis and updating the entire scene with additional training loops, resulting in several hours of processing to edit a single dynamic scene. Therefore, these methods are not scalable with respect to the temporal dimension of the dynamic scene (i.e., the number of timesteps). In this work, we propose an efficient dynamic scene editing method that is more scalable in terms of temporal dimension. To achieve computational efficiency, we leverage a 4D Gaussian representation that models a 4D dynamic scene by combining static 3D Gaussians with a Hexplane-based deformation field, which handles dynamic information. We then perform editing solely on the static 3D Gaussians, which is the minimal but sufficient component required for visual editing. To resolve the misalignment between the edited 3D Gaussians and the deformation field potentially resulting from the editing process, we additionally conducted a refinement stage using a score distillation mechanism. Extensive editing results demonstrate that our method is efficient, reducing editing time by more than half compared to existing methods, while achieving high editing quality that better follows user instructions.

The introduction of 8-bit floating-point (FP8) computation units in modern AI accelerators has generated significant interest in FP8-based large language model (LLM) inference. Unlike 16-bit floating-point formats, FP8 in deep learning requires a shared scaling factor. Additionally, while E4M3 and E5M2 are well-defined at the individual value level, their scaling and accumulation methods remain unspecified and vary across hardware and software implementations. As a result, FP8 behaves more like a quantization format than a standard numeric representation. In this work, we provide the first comprehensive analysis of FP8 computation and acceleration on two AI accelerators: the NVIDIA H100 and Intel Gaudi 2. Our findings highlight that the Gaudi 2, by leveraging FP8, achieves higher throughput-to-power efficiency during LLM inference, offering valuable insights into the practical implications of FP8 adoption for datacenter-scale LLM serving.

In positron emission tomography (PET), time-of-flight (TOF) information localizes source positions along lines of response. Cherenkov-radiator-integrated microchannel-plate photomultiplier tubes have achieved 30 ps TOF resolution, demonstrating cross-sectional imaging without reconstruction. Such ultrafast TOF detectors would free PET from conventional ring geometries. Therefore, this study aimed at investigating imaging characteristics of a dual-panel PET with ultrafast TOF detectors using Geant4 simulation. Two detector panels ($142 \times 142~mm^2$), which consisted of 5.0 mm-thick bismuth germanate pixelized crystals with a 5.75 mm pitch, were placed face-to-face at a 300 mm distance. Spatial resolution and image noise with various TOF resolutions from 30 to 90 ps were evaluated. Because degraded efficiency may cancel TOF gain in image quality, detection efficiency was also parameterized by reducing coincidence counts. A numerical phantom, which consisted of multi-rod and uniform cylindrical activities (21 MBq in total), was simulated for 600 s. Regarding image reconstruction, results of a backprojection (i.e., no reconstruction) were compared with those of the maximum likelihood expectation maximization (MLEM). The dual-panel PET required a 40 ps TOF resolution to have a similar spatial resolution to that of a non-TOF ring PET (300 mm in diameter) for the same detection efficiency. The TOF resolution should be 30 ps with half the efficiency to maintain a similar image noise to that of the 40 ps TOF with complete efficiency. MLEM provided better imaging performance than backprojection, even at 30 ps TOF. The feasibility of the proposed dual-panel PET with ultrafast TOF detectors was shown., Comment: 8 pages, 9 figures

Freezing of gait (FOG) is a debilitating feature of Parkinson's disease (PD), which is a cause of injurious falls among PD patients. Recent advances in wearable-based human activity recognition (HAR) technology have enabled the detection of FOG subtypes across benchmark datasets. Since FOG manifestation is heterogeneous, developing models that quantify FOG consistently across patients with varying demographics, FOG types, and PD conditions is important. Bias and fairness in FOG models remain understudied in HAR, with research focused mainly on FOG detection using single benchmark datasets. We evaluated the bias and fairness of HAR models for wearable-based FOG detection across demographics and PD conditions using multiple datasets and the effectiveness of transfer learning as a potential bias mitigation approach. Our evaluation using demographic parity ratio (DPR) and equalized odds ratio (EOR) showed model bias (DPR & EOR < 0.8) for all stratified demographic variables, including age, sex, and disease duration. Our experiments demonstrated that transfer learning from multi-site datasets and generic human activity representations significantly improved fairness (average change in DPR +0.027, +0.039, respectively) and performance (average change in F1-score +0.026, +0.018, respectively) across attributes, supporting the hypothesis that generic human activity representations learn fairer representations applicable to health analytics., Comment: Submitted to IMWUT 2025

Motivated by the recent construction of grey galaxy and Dual Dressed Black Hole solutions in $AdS_5\times S^5$, we present two conjectures relating to the large $N$ entropy of supersymmetric states in ${\cal N}=4$ Yang-Mills theory. Our first conjecture asserts the existence of a large number of supersymmetric states which can be thought of as a non interacting mix of supersymmetric black holes and supersymmetric `gravitons'. It predicts a microcanonical phase diagram of supersymmetric states with eleven distinct phases, and makes a sharp prediction for the supersymmetric entropy (as a function of 5 charges) in each of these phases. The microcanonical version of the superconformal index involves a sum over states - with alternating signs - over a line in 5 parameter charge space. Our second conjecture asserts that this sum is dominated by the point on the line that has the largest supersymmetric entropy. This conjecture predicts a large $N$ formula for the superconformal index as a function of indicial charges, and predicts a microcanonical indicial phase diagram with nine distinct phases. It predicts agreement between the superconformal index and black hole entropy in one phase (so over one range of charges), but disagreement in other phases (and so at other values of charges). We compare our predictions against numerically evaluated superconformal index at $N\leq10$, and find qualitative agreement., Comment: 59 pages + Appendices, 34 figures

The Quantum Approximate Optimisation Algorithm (QAOA) is a hybrid quantum-classical algorithm for solving combinatorial optimisation problems. QAOA encodes solutions into the ground state of a Hamiltonian, approximated by a $p$-level parameterised quantum circuit composed of problem and mixer Hamiltonians, with parameters optimised classically. While deeper QAOA circuits can offer greater accuracy, practical applications are constrained by complex parameter optimisation and physical limitations such as gate noise, restricted qubit connectivity, and state-preparation-and-measurement errors, limiting implementations to shallow depths. This work focuses on QAOA$_1$ (QAOA at $p=1$) for QUBO problems, represented as Ising models. Despite QAOA$_1$ having only two parameters, $(\gamma, \beta)$, we show that their optimisation is challenging due to a highly oscillatory landscape, with oscillation rates increasing with the problem size, density, and weight. This behaviour necessitates high-resolution grid searches to avoid distortion of cost landscapes that may result in inaccurate minima. We propose an efficient optimisation strategy that reduces the two-dimensional $(\gamma, \beta)$ search to a one-dimensional search over $\gamma$, with $\beta^*$ computed analytically. We establish the maximum permissible sampling period required to accurately map the $\gamma$ landscape and provide an algorithm to estimate the optimal parameters in polynomial time. Furthermore, we rigorously prove that for regular graphs on average, the globally optimal $\gamma^* \in \mathbb{R}^+$ values are concentrated very close to zero and coincide with the first local optimum, enabling gradient descent to replace exhaustive line searches. This approach is validated using Recursive QAOA (RQAOA), where it consistently outperforms both coarsely optimised RQAOA and semidefinite programs across all tested QUBO instances., Comment: 54 pages, 7 Figures, Comments Welcome

The transmon, known for its fast operation time and the coherence time of tens of microseconds, is the most commonly used qubit for superconducting quantum processors. However, it is still necessary to enhance the coherence time and the gate fidelity of superconducting quantum processors for the practical implementation of fault-tolerant quantum computing. Meanwhile, a novel superconducting qubit, which has the ability to protect the Cooper-pair parity on the superconducting island, has been proposed. This new qubit shows better coherence performance than the transmon, but it does not yet have an efficient method for realizing a superconducting hybrid system that harnesses it. In this work, we show how to implement a new type of CNOT gate in a superconducting hybrid system composed of tunable transmon and parity-protected qubit by applying a cross-resonance pulse. First, we provide hardware specifications and pulse parameters to construct a successful two-qubit gate in the hybrid system. Second, we show that our method can supply a CNOT gate of average fidelity with more than 0.998. Therefore, our work implies that the hybrid system may provide a new platform for quantum computers.

This paper is concerned with a compressible MHD equations describing the evolution of viscous non-resistive fluids in piecewise regular bounded Lipschitz domains. Under the general inflow-outflow boundary conditions, we prove existence of global-in-time weak solutions with finite energy initial data. The present result extends considerably the previous work by Li and Sun [\emph{J. Differential Equations.}, 267 (2019), pp. 3827-3851], where the homogeneous Dirichlet boundary condition for velocity field is treated. The proof leans on the specific mathematical structure of equations and the recently developed theory of open fluid systems. Furthermore, we establish the weak-strong uniqueness principle, namely a weak solution coincides with the strong solution on the lifespan of the latter provided they emanate from the same initial and boundary data. This basic property is expected to be useful in the study of convergence of numerical solutions., Comment: 30 pages

The quantum switch has been widely studied as a prototypical example of indefinite causal order in quantum information processing. However, the potential advantages of utilising more general forms of indefinite causal orders remain largely unexplored. We study higher-order switches, which involve concatenated applications of the quantum switch, and we demonstrate that they provide a strict advantage over the conventional quantum switch in the task of quantum channel distillation. Specifically, we show that higher-order quantum switches enable the probabilistic distillation of any qubit Pauli channel into the identity channel with nonzero probability. This capability contrasts with the conventional quantum switch, which allows only a limited set of Pauli channels to be distilled with nonzero probability. We observe that, counterintuitively, the distillation rate generally increases the noisier the channel is. We fully characterise the asymptotic distillation rates of higher-order superswitches for qubit Pauli channels. Finally, we prove a no-go result for multi-qubit generalisations., Comment: 19 pages, 7 figures

Recent advancements in deep learning have significantly improved performance on computer vision tasks. Previous image classification methods primarily modify model architectures or add features, and they optimize models using cross-entropy loss on class logits. Since they focus on classifying images with considering class labels, these methods may struggle to learn various \emph{aspects} of classes (e.g., natural positions and shape changes). Rethinking the previous approach from a novel view, we propose a multi-aspect knowledge distillation method using Multimodal Large Language Models (MLLMs). Our approach involves: 1) querying Large Language Model with multi-aspect questions relevant to the knowledge we want to transfer to the model, 2) extracting corresponding logits from MLLM, and 3) expanding the model's output dimensions to distill these multi-aspect logits. We then apply cross-entropy loss to class logits and binary cross-entropy loss to multi-aspect logits. Through our method, the model can learn not only the knowledge about visual aspects but also the abstract and complex aspects that require a deeper understanding. We primarily apply our method to image classification, and to explore the potential for extending our model, we expand it to other tasks, such as object detection. In all experimental results, our method improves the performance of the baselines. Additionally, we analyze the effect of multi-aspect knowledge distillation. These results demonstrate that our method can transfer knowledge about various aspects to the model and the aspect knowledge can enhance model performance in computer vision tasks. This paper demonstrates the great potential of multi-aspect knowledge distillation, and we believe it offers a promising direction for future research in computer vision and beyond., Comment: Preprint

Envelope methods perform dimension reduction of predictors or responses in multivariate regression, exploiting the relationship between them to improve estimation efficiency. While most research on envelopes has focused on their estimation properties, certain envelope estimators have been shown to excel at prediction in both low and high dimensions. In this paper, we propose to further improve prediction through envelope-guided regularization (EgReg), a novel method which uses envelope-derived information to guide shrinkage along the principal components (PCs) of the predictor matrix. We situate EgReg among other PC-based regression methods and envelope methods to motivate its development. We show that EgReg delivers lower prediction risk than a closely related non-shrinkage envelope estimator when the number of predictors $p$ and observations $n$ are fixed and in any alignment. In an asymptotic regime where the true intrinsic dimension of the predictors and $n$ diverge proportionally, we find that the limiting prediction risk of the non-shrinkage envelope estimator exhibits a double descent phenomenon and is consistently larger than the limiting risk for EgReg. We compare the prediction performance of EgReg with envelope methods and other PC-based prediction methods in simulations and a real data example, observing improved prediction performance over these alternative approaches in general.

LiDAR is a crucial sensor in autonomous driving, commonly used alongside cameras. By exploiting this camera-LiDAR setup and recent advances in image representation learning, prior studies have shown the promising potential of image-to-LiDAR distillation. These prior arts focus on the designs of their own losses to effectively distill the pre-trained 2D image representations into a 3D model. However, the other parts of the designs have been surprisingly unexplored. We find that fundamental design elements, e.g., the LiDAR coordinate system, quantization according to the existing input interface, and data utilization, are more critical than developing loss functions, which have been overlooked in prior works. In this work, we show that simple fixes to these designs notably outperform existing methods by 16% in 3D semantic segmentation on the nuScenes dataset and 13% in 3D object detection on the KITTI dataset in downstream task performance. We focus on overlooked design choices along the spatial and temporal axes. Spatially, prior work has used cylindrical coordinate and voxel sizes without considering their side effects yielded with a commonly deployed sparse convolution layer input interface, leading to spatial quantization errors in 3D models. Temporally, existing work has avoided cumbersome data curation by discarding unsynced data, limiting the use to only the small portion of data that is temporally synced across sensors. We analyze these effects and propose simple solutions for each overlooked aspect., Comment: Accepted to ACCV2024

We study the category $\mathcal{C}$ generated by extremal weight modules over $U_q(\mathfrak{gl}_{>0})$. We show that $\mathcal{C}$ is a tensor category, and give an explicit description of the socle filtration of tensor product of any two extremal weight modules. This follows from the study of Fock space $\mathcal{F}^\infty \otimes \mathcal{M}$ of infinite level, which has commuting actions of a parabolic $q$-boson algebra and $U_p(\mathfrak{gl}_{>0})$ with $p=-q^{-1}$. It contains a (semisimple) limit of the fermionic Fock space $\mathcal{F}^n$ of level $n$, which has a $q$-analogue of Howe duality often called level-rank duality. To describe the socle filtration of $\mathcal{F}^\infty \otimes \mathcal{M}$, we introduce the notion of a saturated crystal valuation, whose existence was observed for example in the embedding of an extremal weight module into a tensor product of fundamental weight modules of affine type due to Kashiwara and Beck-Nakajima., Comment: 72 pages

The scalar spin chirality (SSC), defined as a triple product of spins, is essential for describing noncoplanar spin structures and understanding chiral physics in magnetic systems. Traditionally, SSC has been discussed primarily in the context of noncoplanar ground-state spin configurations at zero temperature, as collinear spin systems are generally thought to lack SSC. Consequently, whether the SSC can emerge at finite temperatures in spin systems with collinear ground states remains an open question and has yet to be fully understood. In this study, we theoretically demonstrate that thermally excited magnons can induce SSC even in collinear spin systems. By considering 2D ferromagnets on Kagome and honeycomb lattices, we demonstrate that the Dzyaloshinskii-Moriya interactions (DMI) which break the effective time-reversal symmetry in the magnon Hamiltonian can lead to finite SSC at finite temperatures. Using a simple spin model, we show both numerically and analytically that the SSC increases with the magnitude of DMI and temperature. Furthermore, calculations based on realistic material parameters reveal that the magnon-induced SSC can achieve a magnitude comparable to those observed in non-coplanar spin configurations. These findings suggest that SSC plays a significant role even in collinear spin systems, providing new insights into the chiral physics of magnetic materials., Comment: 10 pages, 5 figures

We investigate the physics of the spatially random coupling between a Fermi surface and a vector bosonic field in the presence of the magnetic field. We evaluated the propagators of the fermion and vector as well as the self energy and polarization and their dependence on the magnetic field. We also evaluated the temperature dependence of the Hall angle, no strange-metal behaviour is obtained, but the linear-$T$ resistivity is preserved under the magnetic field., Comment: 25 pages, 15 figures

Solving optimization problems is a highly demanding workload requiring high-performance computing systems. Optimization solvers are usually difficult to parallelize in conventional digital architectures, particularly when stochastic decisions are involved. Recently, analog computing architectures for accelerating stochastic optimization solvers have been presented, but they were limited to academic problems in quadratic polynomial format. Here we present KLIMA, a k-Local In-Memory Accelerator with resistive Content Addressable Memories (CAMs) and Dot-Product Engines (DPEs) to accelerate the solution of high-order industry-relevant optimization problems, in particular Boolean Satisfiability. By co-designing the optimization heuristics and circuit architecture we improve the speed and energy to solution up to 182x compared to the digital state of the art.

Mixed-precision quantization methods have been proposed to reduce model size while minimizing accuracy degradation. However, existing studies require retraining and do not consider the computational overhead and intermediate representations (IR) generated during the compilation process, limiting their application at the compiler level. This computational overhead refers to the runtime latency caused by frequent quantization and dequantization operations during inference. Performing these operations at the individual operator level causes significant runtime delays. To address these issues, we propose QuantuneV2, a compiler-based mixed-precision quantization method designed for practical embedded AI applications. QuantuneV2 performs inference only twice, once before quantization and once after quantization, and operates with a computational complexity of O(n) that increases linearly with the number of model parameters. We also made the sensitivity analysis more stable by using local metrics like weights, activation values, the Signal to Quantization Noise Ratio, and the Mean Squared Error. We also cut down on computational overhead by choosing the best IR and using operator fusion. Experimental results show that QuantuneV2 achieved up to a 10.28 percent improvement in accuracy and a 12.52 percent increase in speed compared to existing methods across five models: ResNet18v1, ResNet50v1, SqueezeNetv1, VGGNet, and MobileNetv2. This demonstrates that QuantuneV2 enhances model performance while maintaining computational efficiency, making it suitable for deployment in embedded AI environments., Comment: 18 pages, 10 figures, Accepted in Future Generation Computer Systems Journal

We propose and numerically demonstrate a reconfigurable patch antenna array that enables simultaneous incident wave sensing and anomalous reflection without prior knowledge of the propagation environment. We acquire anomalous reflection by suppressing parasitic scattering through accurate and efficient optimization of induced load currents and by varying impedances of reactive loads. By mitigating parasitic scattering lobes, we demonstrate the feasibility of accurately detecting the incoming illumination angle via the spatial Fourier transform of the optimized load current distribution, facilitated by tunable reactive loads. This approach eliminates the need for additional RF chains, pre-computed data, or calibration measurements. The developed strategy, which integrates arithmetic load optimization with angle-of-arrival sensing, is applicable to general finite-size arrays., Comment: 9 pages, 7 figures

In Reinforcement Learning from Human Feedback (RLHF), it is crucial to learn suitable reward models from human feedback to align large language models (LLMs) with human intentions. However, human feedback can often be noisy, inconsistent, or biased, especially when evaluating complex responses. Such feedback can lead to misaligned reward signals, potentially causing unintended side effects during the RLHF process. To address these challenges, we explore the use of influence functions to measure the impact of human feedback on the performance of reward models. We propose a compute-efficient approximation method that enables the application of influence functions to LLM-based reward models and large-scale preference datasets. In our experiments, we demonstrate two key applications of influence functions: (1) detecting common forms of labeler bias in human feedback datasets and (2) guiding labelers to refine their strategies to align more closely with expert feedback. By quantifying the impact of human feedback on reward models, we believe that influence functions can enhance feedback interpretability and contribute to scalable oversight in RLHF, helping labelers provide more accurate and consistent feedback. Source code is available at https://github.com/mintaywon/IF_RLHF, Comment: Source code: https://github.com/mintaywon/IF_RLHF