Search

We propose a new framework for creating and easily manipulating 3D models of arbitrary objects using casually captured videos. Our core ingredient is a novel hierarchy deformation model, which captures motions of objects with a tree-structured bones. Our hierarchy system decomposes motions based on the granularity and reveals the correlations between parts without exploiting any prior structural knowledge. We further propose to regularize the bones to be positioned at the basis of motions, centers of parts, sufficiently covering related surfaces of the part. This is achieved by our bone occupancy function, which identifies whether a given 3D point is placed within the bone. Coupling the proposed components, our framework offers several clear advantages: (1) users can obtain animatable 3D models of the arbitrary objects in improved quality from their casual videos, (2) users can manipulate 3D models in an intuitive manner with minimal costs, and (3) users can interactively add or delete control points as necessary. The experimental results demonstrate the efficacy of our framework on diverse instances, in reconstruction quality, interpretability and easier manipulation. Our code is available at https://github.com/subin6/HSNB., Comment: ECCV 2024 accepted

In practical electronic applications, where doping is crucial to exploit large-area two-dimensional (2D) semiconductors, surface charge transfer doping (SCTD) has emerged as a promising strategy to tailor their electrical characteristics. However, impurity scattering caused by resultant ionized dopants, after donating or withdrawing carriers, hinders transport in 2D semiconductor layers, limiting the carrier mobility. Here, we propose a diffusion doping method for chemical vapor deposition (CVD) grown molybdenum disulfide that avoids interference from charged impurities. Selectively inkjet-printed dopants were introduced only on the contact region, allowing excessively donated electrons to diffuse to the channel layer due to the electron density difference. Therefore, diffusion-doped molybdenum disulfide FETs do not have undesirable charged impurities on the channel, exhibiting over two-fold higher field-effect mobility compared with conventional direct-doped ones. Our study paves the way for a new doping strategy that simultaneously suppresses charged impurity scattering and facilitates the tailoring of the SCTD effect.

We present a measurement of $|V_{ub}|$ from a simultaneous study of the charmless semileptonic decays $B^0\to\pi^- \ell^+ \nu_{\ell}$ and $B^+\to\rho^0 \ell^+\nu_{\ell}$, where $\ell = e, \mu$. This measurement uses a data sample of 387 million $B\overline{B}$ meson pairs recorded by the Belle~II detector at the SuperKEKB electron-positron collider between 2019 and 2022. The two decays are reconstructed without identifying the partner $B$ mesons. We simultaneously measure the differential branching fractions of $B^0\to\pi^- \ell^+ \nu_{\ell}$ and $B^+\to\rho^0 \ell^+\nu_{\ell}$ decays as functions of $q^2$ (momentum transfer squared). From these, we obtain total branching fractions $B(B^0\to\pi^- \ell^+ \nu_{\ell}) = (1.516 \pm 0.042 (\mathrm{stat}) \pm 0.059 (\mathrm{syst})) \times 10^{-4}$ and $B(B^+\to\rho^0 \ell^+\nu_{\ell}) = (1.625 \pm 0.079 (\mathrm{stat}) \pm 0.180 (\mathrm{syst})) \times 10^{-4}$. By fitting the measured $B^0\to\pi^- \ell^+ \nu_{\ell}$ partial branching fractions as functions of $q^2$, together with constraints on the non-perturbative hadronic contribution from lattice QCD calculations, we obtain $|V_{ub}|$ = $(3.93 \pm 0.09 \pm 0.13 \pm 0.19) \times 10^{-3}$. Here, the first uncertainty is statistical, the second is systematic, and the third is theoretical.

In this work, we introduce Mask-JEPA, a self-supervised learning framework tailored for mask classification architectures (MCA), to overcome the traditional constraints associated with training segmentation models. Mask-JEPA combines a Joint Embedding Predictive Architecture with MCA to adeptly capture intricate semantics and precise object boundaries. Our approach addresses two critical challenges in self-supervised learning: 1) extracting comprehensive representations for universal image segmentation from a pixel decoder, and 2) effectively training the transformer decoder. The use of the transformer decoder as a predictor within the JEPA framework allows proficient training in universal image segmentation tasks. Through rigorous evaluations on datasets such as ADE20K, Cityscapes and COCO, Mask-JEPA demonstrates not only competitive results but also exceptional adaptability and robustness across various training scenarios. The architecture-agnostic nature of Mask-JEPA further underscores its versatility, allowing seamless adaptation to various mask classification family., Comment: 27 pages, 5 figures

We report measurements of time-dependent $CP$ asymmetries in $B^0 \to K^0_S \pi^0 \gamma$ decays based on a data sample of $(388\pm6)\times10^6$ $B\bar{B}$ events collected at the $\Upsilon(4S)$ resonance with the Belle II detector. The Belle II experiment operates at the SuperKEKB asymmetric-energy $e^+e^-$ collider. We measure decay-time distributions to determine $CP$-violating parameters $S$ and $C$. We determine these parameters for two ranges of $K^0_S \pi^0$ invariant mass: $m(K^0_S \pi^0)\in (0.8, 1.0)$ $GeV/c^2$, which is dominated by $B^0 \to K^{*0} (\to K^0_S \pi^0) \gamma$ decays, and a complementary region $m(K^0_S \pi^0)\in (0.6, 0.8)\cup(1.0, 1.8)$ $GeV/c^2$. Our results have improved precision as compared to previous measurements and are consistent with theory predictions., Comment: 10 pages, 4 figures

We present measurements of $B^{+}\rightarrow\rho^{+}\gamma$ and $B^{0}\rightarrow\rho^{0}\gamma$ decays using a combined data sample of $772 \times 10^6$ $B\overline{B}$ pairs collected by the Belle experiment and $387\times 10^6$ $B\overline{B}$ pairs collected by the Belle II experiment in $e^{+}e^{-}$ collisions at the $\Upsilon (4S)$ resonance. After an optimized selection, a simultaneous fit to the Belle and Belle II data sets yields $114\pm 12$ $B^{+}\rightarrow\rho^{+}\gamma$ and $99\pm 12$ $B^{0}\rightarrow\rho^{0}\gamma$ decays. The measured branching fractions are $(13.1^{+2.0 +1.3}_{-1.9 -1.2})\times 10^{-7}$ and $(7.5\pm 1.3^{+1.0}_{-0.8})\times 10^{-7}$ for $B^{+}\rightarrow\rho^{+}\gamma$ and $B^{0}\rightarrow\rho^{0}\gamma$ decays, respectively, where the first uncertainty is statistical and the second is systematic. We also measure the isospin asymmetry $A_{\rm I}(B\rightarrow\rho\gamma)=(10.9^{+11.2 +7.8}_{-11.7 -7.3})\%$ and the direct CP asymmetry $A_{CP}(B^{+}\rightarrow\rho^{+}\gamma)=(-8.2\pm 15.2^{+1.6}_{-1.2})\%$., Comment: 12 pages, 4 figures

We present a search for the baryon number $B$ and lepton number $L$ violating decays $\tau^- \rightarrow \Lambda \pi^-$ and $\tau^- \rightarrow \bar{\Lambda} \pi^-$ produced from the $e^+e^-\to \tau^+\tau^-$ process, using a 364 fb$^{-1}$ data sample collected by the Belle~II experiment at the SuperKEKB collider. No evidence of signal is found in either decay mode, which have $|\Delta(B-L)|$ equal to $2$ and $0$, respectively. Upper limits at 90\% credibility level on the branching fractions of $\tau^- \rightarrow \Lambda\pi^-$ and $\tau^- \rightarrow \bar{\Lambda}\pi^-$ are determined to be $4.7 \times 10^{-8}$ and $4.3 \times 10^{-8}$, respectively., Comment: 8 pages, 4 figures

We report the first evidence for the $h_{b}(\text{2P}) \to \Upsilon(\text{1S})\eta$ transition with a significance of $3.5$ standard deviations. The decay branching fraction is measured to be $\mathcal{B}[h_{b}(\text{2P}) \to \Upsilon(\text{1S})\eta]=(7.1 ~^{+3.7} _{-3.2}\pm 0.8)\times10^{-3}$, which is noticeably smaller than expected. We also set upper limits on $\pi^0$ transitions of $\mathcal{B}[h_{b}(\text{2P}) \to \Upsilon(\text{1S})\pi^0] < 1.8\times10^{-3}$, and $\mathcal{B}[h_{b}(\text{1P})\to \Upsilon(\text{1S})\pi^0] < 1.8\times10^{-3}$, at the $90\%$ confidence level. These results are obtained with a $131.4$~fb$^{-1}$ data sample collected near the $\Upsilon(\text{5S})$ resonance with the Belle detector at the KEKB asymmetric-energy $e^+e^-$ collider., Comment: to be submitted to PRL

Magnetic anisotropy in atomically thin correlated heterostructures is essential for exploring quantum magnetic phases for next-generation spintronics. Whereas previous studies have mostly focused on van der Waals systems, here, we investigate the impact of dimensionality of epitaxially-grown correlated oxides down to the monolayer limit on structural, magnetic, and orbital anisotropies. By designing oxide superlattices with a correlated ferromagnetic SrRuO3 and nonmagnetic SrTiO3 layers, we observed modulated ferromagnetic behavior with the change of the SrRuO3 thickness. Especially, for three-unit-cell-thick layers, we observe a significant 1,500% improvement of coercive field in the anomalous Hall effect, which cannot be solely attributed to the dimensional crossover in ferromagnetism. The atomic-scale heterostructures further reveal the systematic modulation of anisotropy for the lattice structure and orbital hybridization, explaining the enhanced magnetic anisotropy. Our findings provide valuable insights into engineering the anisotropic hybridization of synthetic magnetic crystals, offering a tunable spin order for various applications., Comment: 23 pages

Interactions with billion-scale large language models typically yield long-form responses due to their extensive parametric capacities, along with retrieval-augmented features. While detailed responses provide insightful viewpoint of a specific subject, they frequently generate redundant and less engaging content that does not meet user interests. In this work, we focus on the role of query outlining (i.e., selected sequence of queries) in scenarios that users request a specific range of information, namely coverage-conditioned ($C^2$) scenarios. For simulating $C^2$ scenarios, we construct QTree, 10K sets of information-seeking queries decomposed with various perspectives on certain topics. By utilizing QTree, we train QPlanner, a 7B language model generating customized query outlines that follow coverage-conditioned queries. We analyze the effectiveness of generated outlines through automatic and human evaluation, targeting on retrieval-augmented generation (RAG). Moreover, the experimental results demonstrate that QPlanner with alignment training can further provide outlines satisfying diverse user interests. Our resources are available at https://github.com/youngerous/qtree., Comment: Work in progress. Resources are available at https://github.com/youngerous/qtree

A series of data samples was collected with the Belle II detector at the SuperKEKB collider from March 2019 to June 2022. We determine the integrated luminosities of these data samples using three distinct methodologies involving Bhabha ($e^+e^- \to e^+e^-(n\gamma)$), digamma ($e^+e^- \to \gamma\gamma(n\gamma)$), and dimuon ($e^+e^- \to \mu^+ \mu^- (n\gamma)$) events. The total integrated luminosity obtained with Bhabha, digamma, and dimuon events is (426.52 $\pm$ 0.03 $\pm$ 2.48)~fb$^{-1}$, (427.32 $\pm$ 0.03 $\pm$ 2.56)~fb$^{-1}$, and (424.84 $\pm$ 0.04 $\pm$ 3.88)~fb$^{-1}$, where the first uncertainties are statistical and the second are systematic. The resulting total integrated luminosity obtained from the combination of the three methods is (426.88 $\pm$ 1.93)~fb$^{-1}$., Comment: 12 pages, 3 figures

Existing datasets for 3D hand-object interaction are limited either in the data cardinality, data variations in interaction scenarios, or the quality of annotations. In this work, we present a comprehensive new training dataset for hand-object interaction called HOGraspNet. It is the only real dataset that captures full grasp taxonomies, providing grasp annotation and wide intraclass variations. Using grasp taxonomies as atomic actions, their space and time combinatorial can represent complex hand activities around objects. We select 22 rigid objects from the YCB dataset and 8 other compound objects using shape and size taxonomies, ensuring coverage of all hand grasp configurations. The dataset includes diverse hand shapes from 99 participants aged 10 to 74, continuous video frames, and a 1.5M RGB-Depth of sparse frames with annotations. It offers labels for 3D hand and object meshes, 3D keypoints, contact maps, and \emph{grasp labels}. Accurate hand and object 3D meshes are obtained by fitting the hand parametric model (MANO) and the hand implicit function (HALO) to multi-view RGBD frames, with the MoCap system only for objects. Note that HALO fitting does not require any parameter tuning, enabling scalability to the dataset's size with comparable accuracy to MANO. We evaluate HOGraspNet on relevant tasks: grasp classification and 3D hand pose estimation. The result shows performance variations based on grasp type and object class, indicating the potential importance of the interaction space captured by our dataset. The provided data aims at learning universal shape priors or foundation models for 3D hand-object interaction. Our dataset and code are available at https://hograspnet2024.github.io/., Comment: 14 pages except for references. It will be published at European Conference on Computer Vision(ECCV) 2024

In the current landscape of autonomous vehicle (AV) safety and security research, there are multiple isolated problems being tackled by the community at large. Due to the lack of common evaluation criteria, several important research questions are at odds with one another. For instance, while much research has been conducted on physical attacks deceiving AV perception systems, there is often inadequate investigations on working defenses and on the downstream effects of safe vehicle control. This paper provides a thorough description of the current state of AV safety and security research. We provide individual sections for the primary research questions that concern this research area, including AV surveillance, sensor system reliability, security of the AV stack, algorithmic robustness, and safe environment interaction. We wrap up the paper with a discussion of the issues that concern the interactions of these separate problems. At the conclusion of each section, we propose future research questions that still lack conclusive answers. This position article will serve as an entry point to novice and veteran researchers seeking to partake in this research domain., Comment: Accepted to 1st Cyber Security in Cars Workshop (CSCS) at CCS

Large Language Models (LLMs) have demonstrated remarkable proficiency in various natural language generation (NLG) tasks. Previous studies suggest that LLMs' generation process involves uncertainty. However, existing approaches to uncertainty estimation mainly focus on sequence-level uncertainty, overlooking individual pieces of information within sequences. These methods fall short in separately assessing the uncertainty of each component in a sequence. In response, we propose a novel framework for Concept-Level Uncertainty Estimation (CLUE) for LLMs. We leverage LLMs to convert output sequences into concept-level representations, breaking down sequences into individual concepts and measuring the uncertainty of each concept separately. We conduct experiments to demonstrate that CLUE can provide more interpretable uncertainty estimation results compared with sentence-level uncertainty, and could be a useful tool for various tasks such as hallucination detection and story generation.

The EXperiment for Cryogenic Large-Aperture Intensity Mapping (EXCLAIM) is a balloon-borne telescope designed to survey star formation over cosmological time scales using intensity mapping in the 420 - 540 GHz frequency range. EXCLAIM uses a fully cryogenic telescope coupled to six on-chip spectrometers featuring kinetic inductance detectors (KIDs) to achieve high sensitivity, allowing for fast integration in dark atmospheric windows. The telescope receiver is cooled to $\approx$ 1.7 K by immersion in a superfluid helium bath and enclosed in a superfluid-tight shell with a meta-material anti-reflection coated silicon window. In addition to the optics and the spectrometer package, the receiver contains the magnetic shielding, the cryogenic segment of the spectrometer readout, and the sub-Kelvin cooling system. A three-stage continuous adiabatic demagnetization refrigerator (CADR) keeps the detectors at 100 mK while a $^4$He sorption cooler provides a 900 mK thermal intercept for mechanical suspensions and coaxial cables. We present the design of the EXCLAIM receiver and report on the flight-like testing of major receiver components, including the superfluid-tight receiver window and the sub-Kelvin coolers., Comment: 13 pages, 7 figures, 1 table

Transfer learning based on full fine-tuning (FFT) of the pre-trained encoder and task-specific decoder becomes increasingly complex as deep models grow exponentially. Parameter efficient fine-tuning (PEFT) approaches using adapters consisting of small learnable layers have emerged as an alternative to FFT, achieving comparable performance while maintaining high training efficiency. However, the inflexibility of the adapter with respect to input instances limits its capability of learning task-specific information in diverse downstream tasks. In this paper, we propose a novel PEFT approach, input-Conditioned transFormer, termed iConFormer, that leverages a dynamic adapter conditioned on the input instances. To secure flexible learning ability on input instances in various downstream tasks, we introduce an input-Conditioned Network (iCoN) in the dynamic adapter that enables instance-level feature transformation. To be specific, iCoN generates channel-wise convolutional kernels for each feature and transform it using adaptive convolution process to effectively capture task-specific and fine-grained details tailor to downstream tasks. Experimental results demonstrate that by tuning just 1.6% to 2.8% of the Transformer backbone parameters, iConFormer achieves performance comparable to FFT in monocular depth estimation and semantic segmentation, while outperforming it in image classification and instance segmentation. Also, the proposed method consistently outperforms recent PEFT methods for all the tasks mentioned above.

Van der Waals (vdW) magnets have opened a new avenue of novel opportunities covering various interesting phases. Co1/3TaS2-an intercalated metallic vdW antiferromagnet-is one of the latest important additions to the growing list of materials due to its unique triple-Q (3Q) ground state possessing topological characteristics. Careful bulk characterisations have shown the ground state of CoxTaS2 to be a rare 3Q tetrahedral structure for x less than 1/3. The uniqueness of this ground state arises from the dense real-space Berry curvature due to scalar spin chirality, giving rise to a noticeable anomalous Hall effect. In this work, we demonstrate that we can control this topological phase via gating. Using three kinds of CoxTaS2 devices with different Co compositions, we have established that we can cover the whole 3Q topological phase with ionic gating. This work reports a rare demonstration of electrical gating control of layered antiferromagnetic metal. More importantly, our work constitutes one of the first examples of the electrical control of the scalar spin chirality using antiferromagnetic metal., Comment: 19 pages, 4 figures

Unsupervised Domain Adaptation (UDA) endeavors to bridge the gap between a model trained on a labeled source domain and its deployment in an unlabeled target domain. However, current high-performance models demand significant resources, resulting in prohibitive deployment costs and highlighting the need for small yet effective models. For UDA of lightweight models, Knowledge Distillation (KD) in a Teacher-Student framework can be a common approach, but we find that domain shift in UDA leads to a significant increase in non-salient parameters in the teacher model, degrading model's generalization ability and transferring misleading information to the student model. Interestingly, we observed that this phenomenon occurs considerably less in the student model. Driven by this insight, we introduce Collaborative Learning, a method that updates the teacher's non-salient parameters using the student model and at the same time enhance the student's performance using the updated teacher model. Experiments across various tasks and datasets show consistent performance improvements for both student and teacher models. For example, in semantic segmentation, CLDA achieves an improvement of +0.7% mIoU for teacher and +1.4% mIoU for student compared to the baseline model in the GTA to Cityscapes. In the Synthia to Cityscapes, it achieves an improvement of +0.8% mIoU for teacher and +2.0% mIoU for student.

We analyses occurrence of DH type II solar radio bursts spanning over solar cycles 23-25 during which a total of 590 DH type II bursts are reported with confirmed 568 and 462 cases of associated CME and flares, respectively. We find short-term yet important differences in DH type II activity when the data is examined in terms of event counts and their durations, e.g., temporal shift in the peak activity during cycle 24 and variation in the growth rate of the activity level during cycle 25. For an in-depth exploration, DH type II bursts are classified in 3 categories based on their end-frequencies: Low-, Medium-, and High- Frequency Groups (LFG, MFG, and HFG, respectively). The HFG category is the most populous (~47 %) while the LFG category occupy about a quarter of the events (~24 %). The LFG events show a clear inclination toward fastest CMEs and X-class flares with a quarter of events exhibiting end frequency below 50 MHz., Comment: 6 pages, 6 figures, 4 tables, Proceedings IAU Symposium No. 388

Indoor robots rely on depth to perform tasks like navigation or obstacle detection, and single-image depth estimation is widely used to assist perception. Most indoor single-image depth prediction focuses less on model generalizability to unseen datasets, concerned with in-the-wild robustness for system deployment. This work leverages gradient-based meta-learning to gain higher generalizability on zero-shot cross-dataset inference. Unlike the most-studied meta-learning of image classification associated with explicit class labels, no explicit task boundaries exist for continuous depth values tied to highly varying indoor environments regarding object arrangement and scene composition. We propose fine-grained task that treats each RGB-D mini-batch as a task in our meta-learning formulation. We first show that our method on limited data induces a much better prior (max 27.8% in RMSE). Then, finetuning on meta-learned initialization consistently outperforms baselines without the meta approach. Aiming at generalization, we propose zero-shot cross-dataset protocols and validate higher generalizability induced by our meta-initialization, as a simple and useful plugin to many existing depth estimation methods. The work at the intersection of depth and meta-learning potentially drives both research to step closer to practical robotic and machine perception usage., Comment: IROS 2024. The version supersedes 2305.07269. arXiv admin note: text overlap with arXiv:2305.07269

Articulatory trajectories like electromagnetic articulography (EMA) provide a low-dimensional representation of the vocal tract filter and have been used as natural, grounded features for speech synthesis. Differentiable digital signal processing (DDSP) is a parameter-efficient framework for audio synthesis. Therefore, integrating low-dimensional EMA features with DDSP can significantly enhance the computational efficiency of speech synthesis. In this paper, we propose a fast, high-quality, and parameter-efficient DDSP articulatory vocoder that can synthesize speech from EMA, F0, and loudness. We incorporate several techniques to solve the harmonics / noise imbalance problem, and add a multi-resolution adversarial loss for better synthesis quality. Our model achieves a transcription word error rate (WER) of 6.67% and a mean opinion score (MOS) of 3.74, with an improvement of 1.63% and 0.16 compared to the state-of-the-art (SOTA) baseline. Our DDSP vocoder is 4.9x faster than the baseline on CPU during inference, and can generate speech of comparable quality with only 0.4M parameters, in contrast to the 9M parameters required by the SOTA., Comment: accepted for Spoken Language Technology Workshop 2024

A force field as accurate as quantum mechanics (QM) and as fast as molecular mechanics (MM), with which one can simulate a biomolecular system efficiently enough and meaningfully enough to get quantitative insights, is among the most ardent dreams of biophysicists -- a dream, nevertheless, not to be fulfilled any time soon. Machine learning force fields (MLFFs) represent a meaningful endeavor towards this direction, where differentiable neural functions are parametrized to fit ab initio energies, and furthermore forces through automatic differentiation. We argue that, as of now, the utility of the MLFF models is no longer bottlenecked by accuracy but primarily by their speed (as well as stability and generalizability), as many recent variants, on limited chemical spaces, have long surpassed the chemical accuracy of $1$ kcal/mol -- the empirical threshold beyond which realistic chemical predictions are possible -- though still magnitudes slower than MM. Hoping to kindle explorations and designs of faster, albeit perhaps slightly less accurate MLFFs, in this review, we focus our attention on the design space (the speed-accuracy tradeoff) between MM and ML force fields. After a brief review of the building blocks of force fields of either kind, we discuss the desired properties and challenges now faced by the force field development community, survey the efforts to make MM force fields more accurate and ML force fields faster, envision what the next generation of MLFF might look like.

Optimization over Stiefel manifolds has found wide applications in many scientific and engineering domains. Despite considerable research effort, high-dimensional optimization problems over Stiefel manifolds remain challenging, and the situation is exacerbated by nonsmooth objective functions. The purpose of this paper is to propose and study a novel coordinate-type algorithm for weakly convex (possibly nonsmooth) optimization problems over high-dimensional Stiefel manifolds, named randomized submanifold subgradient method (RSSM). Similar to coordinate-type algorithms in the Euclidean setting, RSSM exhibits low per-iteration cost and is suitable for high-dimensional problems. We prove that RSSM converges to the set of stationary points and attains $\varepsilon$-stationary points with respect to a natural stationarity measure in $\mathcal{O}(\varepsilon^{-4})$ iterations in both expectation and the almost-sure senses. To the best of our knowledge, these are the first convergence guarantees for coordinate-type algorithms to optimize nonconvex nonsmooth functions over Stiefel manifolds. An important technical tool in our convergence analysis is a new Riemannian subgradient inequality for weakly convex functions on proximally smooth matrix manifolds, which could be of independent interest.

Large language models (LLMs) have emerged due to their capability to generate high-quality content across diverse contexts. To reduce their explosively increasing demands for computing resources, a mixture of experts (MoE) has emerged. The MoE layer enables exploiting a huge number of parameters with less computation. Applying state-of-the-art continuous batching increases throughput; however, it leads to frequent DRAM access in the MoE and attention layers. We observe that conventional computing devices have limitations when processing the MoE and attention layers, which dominate the total execution time and exhibit low arithmetic intensity (Op/B). Processing MoE layers only with devices targeting low-Op/B such as processing-in-memory (PIM) architectures is challenging due to the fluctuating Op/B in the MoE layer caused by continuous batching. To address these challenges, we propose Duplex, which comprises xPU tailored for high-Op/B and Logic-PIM to effectively perform low-Op/B operation within a single device. Duplex selects the most suitable processor based on the Op/B of each layer within LLMs. As the Op/B of the MoE layer is at least 1 and that of the attention layer has a value of 4-8 for grouped query attention, prior PIM architectures are not efficient, which place processing units inside DRAM dies and only target extremely low-Op/B (under one) operations. Based on recent trends, Logic-PIM adds more through-silicon vias (TSVs) to enable high-bandwidth communication between the DRAM die and the logic die and place powerful processing units on the logic die, which is best suited for handling low-Op/B operations ranging from few to a few dozens. To maximally utilize the xPU and Logic-PIM, we propose expert and attention co-processing., Comment: 15 pages, 16 figures, accepted at MICRO 2024

Recent research has indicated a substantial rise in interest in understanding Nesterov's accelerated gradient methods via their continuous-time models. However, most existing studies focus on specific classes of Nesterov's methods, which hinders the attainment of an in-depth understanding and a unified perspective. To address this deficit, we present generalized continuous-time models that cover a broad range of Nesterov's methods, including those previously studied under existing continuous-time frameworks. Our key contributions are as follows. First, we identify the convergence rates of the generalized models, eliminating the need to determine the convergence rate for any specific continuous-time model derived from them. Second, we show that six existing continuous-time models are special cases of our generalized models, thereby positioning our framework as a unifying tool for analyzing and understanding these models. Third, we design a restart scheme for Nesterov's methods based on our generalized models and show that it ensures a monotonic decrease in objective function values. Owing to the broad applicability of our models, this scheme can be used to a broader class of Nesterov's methods compared to the original restart scheme. Fourth, we uncover a connection between our generalized models and gradient flow in continuous time, showing that the accelerated convergence rates of our generalized models can be attributed to a time reparametrization in gradient flow. Numerical experiment results are provided to support our theoretical analyses and results.

We introduce SonoHaptics, an audio-haptic cursor for gaze-based 3D object selection. SonoHaptics addresses challenges around providing accurate visual feedback during gaze-based selection in Extended Reality (XR), e.g., lack of world-locked displays in no- or limited-display smart glasses and visual inconsistencies. To enable users to distinguish objects without visual feedback, SonoHaptics employs the concept of cross-modal correspondence in human perception to map visual features of objects (color, size, position, material) to audio-haptic properties (pitch, amplitude, direction, timbre). We contribute data-driven models for determining cross-modal mappings of visual features to audio and haptic features, and a computational approach to automatically generate audio-haptic feedback for objects in the user's environment. SonoHaptics provides global feedback that is unique to each object in the scene, and local feedback to amplify differences between nearby objects. Our comparative evaluation shows that SonoHaptics enables accurate object identification and selection in a cluttered scene without visual feedback., Comment: UIST 2024

Event detection and text reasoning have become critical applications across various domains. While LLMs have recently demonstrated impressive progress in reasoning abilities, they often struggle with event detection, particularly due to the absence of training methods that consider causal relationships between event triggers and types. To address this challenge, we propose a novel approach for instruction fine-tuning LLMs for event detection. Our method introduces Semantic Causal Graphs (SCGs) to capture both causal relationships and contextual information within text. Building off of SCGs, we propose SCG Instructions for fine-tuning LLMs by focusing on event triggers and their relationships to event types, and employ Low-Rank Adaptation (LoRA) to help preserve the general reasoning abilities of LLMs. Our evaluations demonstrate that training LLMs with SCG Instructions outperforms standard instruction fine-tuning by an average of 35.69\% on Event Trigger Classification. Notably, our fine-tuned Mistral 7B model also outperforms GPT-4 on key event detection metrics by an average of 31.01\% on Event Trigger Identification, 37.40\% on Event Trigger Classification, and 16.43\% on Event Classification. We analyze the retention of general capabilities, observing only a minimal average drop of 2.03 points across six benchmarks. This comprehensive study investigates multiple LLMs for the event detection task across various datasets, prompting strategies, and training approaches.

The iron-based superconductor FeTe$_{0.55}$Se$_{0.45}$ is a peculiar material: it hosts a surface state with a Dirac dispersion, is a putative topological superconductor hosting Majorana modes in vortices, and has an unusually low Fermi energy. The superconducting state is generally thought to be characterized by three gaps in different bands, with the usual homogenous, spatially extended Bogoliubov excitations -- in this work, we uncover evidence that it is instead of a very different nature. Our scanning tunneling spectroscopy data shows several peaks in the density of states above a full gap, and by analyzing the spatial and junction-resistance dependence of the peaks, we conclude that the peaks above the first one are not coherence peaks from different bands. Instead, comparisons with our simulations indicate that they originate from generalized Shiba states that are spatially overlapping. This can lead to an amorphous state of Bogoliubov quasiparticles, reminiscent of impurity bands in semiconductors. We discuss the origin and implications of this new state., Comment: 6 pages, 4 figures

Temporal Action Detection (TAD) is fundamental yet challenging for real-world video applications. Leveraging the unique benefits of transformers, various DETR-based approaches have been adopted in TAD. However, it has recently been identified that the attention collapse in self-attention causes the performance degradation of DETR for TAD. Building upon previous research, this paper newly addresses the attention collapse problem in cross-attention within DETR-based TAD methods. Moreover, our findings reveal that cross-attention exhibits patterns distinct from predictions, indicating a short-cut phenomenon. To resolve this, we propose a new framework, Prediction-Feedback DETR (Pred-DETR), which utilizes predictions to restore the collapse and align the cross- and self-attention with predictions. Specifically, we devise novel prediction-feedback objectives using guidance from the relations of the predictions. As a result, Pred-DETR significantly alleviates the collapse and achieves state-of-the-art performance among DETR-based methods on various challenging benchmarks including THUMOS14, ActivityNet-v1.3, HACS, and FineAction.

This paper examines the Code-Switching (CS) phenomenon where two languages intertwine within a single utterance. There exists a noticeable need for research on the CS between English and Korean. We highlight that the current Equivalence Constraint (EC) theory for CS in other languages may only partially capture English-Korean CS complexities due to the intrinsic grammatical differences between the languages. We introduce a novel Koglish dataset tailored for English-Korean CS scenarios to mitigate such challenges. First, we constructed the Koglish-GLUE dataset to demonstrate the importance and need for CS datasets in various tasks. We found the differential outcomes of various foundation multilingual language models when trained on a monolingual versus a CS dataset. Motivated by this, we hypothesized that SimCSE, which has shown strengths in monolingual sentence embedding, would have limitations in CS scenarios. We construct a novel Koglish-NLI (Natural Language Inference) dataset using a CS augmentation-based approach to verify this. From this CS-augmented dataset Koglish-NLI, we propose a unified contrastive learning and augmentation method for code-switched embeddings, ConCSE, highlighting the semantics of CS sentences. Experimental results validate the proposed ConCSE with an average performance enhancement of 1.77\% on the Koglish-STS(Semantic Textual Similarity) tasks., Comment: ICPR 2024

Reinforcement learning (RL) with diverse offline datasets can have the advantage of leveraging the relation of multiple tasks and the common skills learned across those tasks, hence allowing us to deal with real-world complex problems efficiently in a data-driven way. In offline RL where only offline data is used and online interaction with the environment is restricted, it is yet difficult to achieve the optimal policy for multiple tasks, especially when the data quality varies for the tasks. In this paper, we present a skill-based multi-task RL technique on heterogeneous datasets that are generated by behavior policies of different quality. To learn the shareable knowledge across those datasets effectively, we employ a task decomposition method for which common skills are jointly learned and used as guidance to reformulate a task in shared and achievable subtasks. In this joint learning, we use Wasserstein auto-encoder (WAE) to represent both skills and tasks on the same latent space and use the quality-weighted loss as a regularization term to induce tasks to be decomposed into subtasks that are more consistent with high-quality skills than others. To improve the performance of offline RL agents learned on the latent space, we also augment datasets with imaginary trajectories relevant to high-quality skills for each task. Through experiments, we show that our multi-task offline RL approach is robust to the mixed configurations of different-quality datasets and it outperforms other state-of-the-art algorithms for several robotic manipulation tasks and drone navigation tasks., Comment: 12 pages, 5 figures, acceepted in NeurIPS 2022

Vertical Federated Learning (VFL) focuses on handling vertically partitioned data over FL participants. Recent studies have discovered a significant vulnerability in VFL to backdoor attacks which specifically target the distinct characteristics of VFL. Therefore, these attacks may neutralize existing defense mechanisms designed primarily for Horizontal Federated Learning (HFL) and deep neural networks. In this paper, we present the first backdoor defense, called VFLIP, specialized for VFL. VFLIP employs the identification and purification techniques that operate at the inference stage, consequently improving the robustness against backdoor attacks to a great extent. VFLIP first identifies backdoor-triggered embeddings by adopting a participant-wise anomaly detection approach. Subsequently, VFLIP conducts purification which removes the embeddings identified as malicious and reconstructs all the embeddings based on the remaining embeddings. We conduct extensive experiments on CIFAR10, CINIC10, Imagenette, NUS-WIDE, and BankMarketing to demonstrate that VFLIP can effectively mitigate backdoor attacks in VFL. https://github.com/blingcho/VFLIP-esorics24, Comment: Accepted by 29th European Symposium on Research in Computer Security (ESORICS 2024)

We propose a novel machine learning approach for inferring causal variables of a target variable from observations. Our goal is to identify both direct and indirect causes within a system, thereby efficiently regulating the target variable when the difficulty and cost of intervening on each causal variable vary. Our method employs a neural network trained to identify causality through supervised learning on simulated data. By implementing a local-inference strategy, we achieve linear complexity with respect to the number of variables, efficiently scaling up to thousands of variables. Empirical results demonstrate the effectiveness of our method in identifying causal relationships within large-scale gene regulatory networks, outperforming existing causal discovery methods that primarily focus on direct causality. We validate our model's generalization capability across novel graph structures and generating mechanisms, including gene regulatory networks of E. coli and the human K562 cell line. Implementation codes are available at https://github.com/snu-mllab/Targeted-Cause-Discovery., Comment: preprint

A Bass order is an order of a number field whose fractional ideals are generated by two elements. Majority of number fields contain infinitely many Bass orders. For example, any order of a number field which contains the maximal order of a subfield with degree 2 or whose discriminant is 4th-power-free in $\mathbb{Z}$, is a Bass order. In this paper, we will propose a closed formula for the number of fractional ideals of a Bass order $R$, up to its invertible ideals, using the conductor of $R$. We will also explain explicit enumeration of all orders containing $R$. Our method is based on local global argument and exhaustion argument, by using orbital integrals for $\mathfrak{gl}_n$ as a mass formula., Comment: 50 pages

Mid-air navigation offers a method of aerial travel that mitigates the constraints associated with continuous navigation. A mid-air selection technique is essential to enable such navigation. In this paper, we consider four variations of intersection-based bimanual mid-air selection techniques with visual aids and supporting features: Simple-Ray, Simple-Stripe, Precision-Stripe, and Cursor-Sync. We evaluate their performance and user experience compared to an unimanual mid-air selection technique using two tasks that require selecting a mid-air position with or without a reference object. Our findings indicate that the bimanual techniques generally demonstrate faster selection times compared to the unimanual technique. With a supporting feature, the bimanual techniques can provide a more accurate selection than the unimanual technique. Based on our results, we discuss the effect of selection technique's visual aids and supporting features on performance and user experience for mid-air selection., Comment: This paper is accepted as a conference paper on ISMAR 2024

Two celebrated extensions of the classical Helly's theorem are the fractional Helly theorem and the colorful Helly theorem. Bulavka, Goodarzi, and Tancer recently established the optimal bound for the unified generalization of the fractional and the colorful Helly theorems using a colored extension of the exterior algebra. In this paper, we combinatorially reduce both the fractional Helly theorem and its colorful version to a classical problem in extremal combinatorics known as {weak saturation}. No such results connecting the fractional Helly theorem and weak saturation are known in the long history of literature. These reductions, along with basic linear algebraic arguments for the reduced weak saturation problems, let us give new short proofs of the optimal bounds for both the fractional Helly theorem and its colorful version without using exterior algebra., Comment: 5 pages, 1 figure

Video deblurring aims to enhance the quality of restored results in motion-blurred videos by effectively gathering information from adjacent video frames to compensate for the insufficient data in a single blurred frame. However, when faced with consecutively severe motion blur situations, frame-based video deblurring methods often fail to find accurate temporal correspondence among neighboring video frames, leading to diminished performance. To address this limitation, we aim to solve the video deblurring task by leveraging an event camera with micro-second temporal resolution. To fully exploit the dense temporal resolution of the event camera, we propose two modules: 1) Intra-frame feature enhancement operates within the exposure time of a single blurred frame, iteratively enhancing cross-modality features in a recurrent manner to better utilize the rich temporal information of events, 2) Inter-frame temporal feature alignment gathers valuable long-range temporal information to target frames, aggregating sharp features leveraging the advantages of the events. In addition, we present a novel dataset composed of real-world blurred RGB videos, corresponding sharp videos, and event data. This dataset serves as a valuable resource for evaluating event-guided deblurring methods. We demonstrate that our proposed methods outperform state-of-the-art frame-based and event-based motion deblurring methods through extensive experiments conducted on both synthetic and real-world deblurring datasets. The code and dataset are available at https://github.com/intelpro/CMTA., Comment: Accepted in ECCV2024

In low-light conditions, capturing videos with frame-based cameras often requires long exposure times, resulting in motion blur and reduced visibility. While frame-based motion deblurring and low-light enhancement have been studied, they still pose significant challenges. Event cameras have emerged as a promising solution for improving image quality in low-light environments and addressing motion blur. They provide two key advantages: capturing scene details well even in low light due to their high dynamic range, and effectively capturing motion information during long exposures due to their high temporal resolution. Despite efforts to tackle low-light enhancement and motion deblurring using event cameras separately, previous work has not addressed both simultaneously. To explore the joint task, we first establish real-world datasets for event-guided low-light enhancement and deblurring using a hybrid camera system based on beam splitters. Subsequently, we introduce an end-to-end framework to effectively handle these tasks. Our framework incorporates a module to efficiently leverage temporal information from events and frames. Furthermore, we propose a module to utilize cross-modal feature information to employ a low-pass filter for noise suppression while enhancing the main structural information. Our proposed method significantly outperforms existing approaches in addressing the joint task. Our project pages are available at https://github.com/intelpro/ELEDNet., Comment: Accepted in ECCV2024

Out-of-distribution (OOD) detection, which determines whether a given sample is part of the in-distribution (ID), has recently shown promising results through training with synthetic OOD datasets. Nonetheless, existing methods often produce outliers that are considerably distant from the ID, showing limited efficacy for capturing subtle distinctions between ID and OOD. To address these issues, we propose a novel framework, Semantic Outlier generation via Nuisance Awareness (SONA), which notably produces challenging outliers by directly leveraging pixel-space ID samples through diffusion models. Our approach incorporates SONA guidance, providing separate control over semantic and nuisance regions of ID samples. Thereby, the generated outliers achieve two crucial properties: (i) they present explicit semantic-discrepant information, while (ii) maintaining various levels of nuisance resemblance with ID. Furthermore, the improved OOD detector training with SONA outliers facilitates learning with a focus on semantic distinctions. Extensive experiments demonstrate the effectiveness of our framework, achieving an impressive AUROC of 88% on near-OOD datasets, which surpasses the performance of baseline methods by a significant margin of approximately 6%.

Diffusion-based video generation models have made significant strides, producing outputs with improved visual fidelity, temporal coherence, and user control. These advancements hold great promise for improving surgical education by enabling more realistic, diverse, and interactive simulation environments. In this study, we introduce SurGen, a text-guided diffusion model tailored for surgical video synthesis, producing the highest resolution and longest duration videos among existing surgical video generation models. We validate the visual and temporal quality of the outputs using standard image and video generation metrics. Additionally, we assess their alignment to the corresponding text prompts through a deep learning classifier trained on surgical data. Our results demonstrate the potential of diffusion models to serve as valuable educational tools for surgical trainees.

COSINE-100 is a direct dark matter search experiment, with the primary goal of testing the annual modulation signal observed by DAMA/LIBRA, using the same target material, NaI(Tl). In previous analyses, we achieved the same 1 keV energy threshold used in the DAMA/LIBRA's analysis that reported an annual modulation signal with 11.6$\sigma$ significance. In this article, we report an improved analysis that lowered the threshold to 0.7 keV, thanks to the application of Multi-Layer Perception network and a new likelihood parameter with waveforms in the frequency domain. The lower threshold would enable a better comparison of COSINE-100 with new DAMA results with a 0.75 keV threshold and account for differences in quenching factors. Furthermore the lower threshold can enhance COSINE-100's sensitivity to sub-GeV dark matter searches.

Materials with chiral electronic structures are of great interest. We report a triazatruxene, TAT, molecular semiconductor with chiral alkyl side chains that crystallises from solution to form chirally-stacked columns with a helical pitch of 6 TATs (2.3 nm). These crystals show strong circularly polarised, CP, green photoluminescence, with dissymmetry of 24%. Electronic structure calculations using the full crystal structure, show that this chiral stacking associates angular momentum to the valence and conduction states and thus gives rise to the observed CP luminescence. Free-standing crystals are not useful for active semiconductor devices, but we have discovered that co-sublimation of TAT as the guest in a structurally mismatched host enables the fabrication of thin films where the chiral crystallization is achieved in-situ by thermally-triggered nano-phase segregation of dopant and host whilst preserving the integrity of the film. This enables fabrication of bright (green) organic light-emitting diodes with unexpectedly high external quantum efficiencies of up to 16% and electroluminescence dissymmetries above 10%. These materials and this process method offer significant application potential in spintronics, optical displays and multidimensional optoelectronics.

Indoor monocular depth estimation helps home automation, including robot navigation or AR/VR for surrounding perception. Most previous methods primarily experiment with the NYUv2 Dataset and concentrate on the overall performance in their evaluation. However, their robustness and generalization to diversely unseen types or categories for indoor spaces (spaces types) have yet to be discovered. Researchers may empirically find degraded performance in a released pretrained model on custom data or less-frequent types. This paper studies the common but easily overlooked factor-space type and realizes a model's performance variances across spaces. We present InSpaceType Dataset, a high-quality RGBD dataset for general indoor scenes, and benchmark 13 recent state-of-the-art methods on InSpaceType. Our examination shows that most of them suffer from performance imbalance between head and tailed types, and some top methods are even more severe. The work reveals and analyzes underlying bias in detail for transparency and robustness. We extend the analysis to a total of 4 datasets and discuss the best practice in synthetic data curation for training indoor monocular depth. Further, dataset ablation is conducted to find out the key factor in generalization. This work marks the first in-depth investigation of performance variances across space types and, more importantly, releases useful tools, including datasets and codes, to closely examine your pretrained depth models. Data and code: https://depthcomputation.github.io/DepthPublic/, Comment: BMVC 2024. This version supersedes 2309.13516

We conducted an experiment during the review process of the 2023 International Conference on Machine Learning (ICML) that requested authors with multiple submissions to rank their own papers based on perceived quality. We received 1,342 rankings, each from a distinct author, pertaining to 2,592 submissions. In this paper, we present an empirical analysis of how author-provided rankings could be leveraged to improve peer review processes at machine learning conferences. We focus on the Isotonic Mechanism, which calibrates raw review scores using author-provided rankings. Our analysis demonstrates that the ranking-calibrated scores outperform raw scores in estimating the ground truth ``expected review scores'' in both squared and absolute error metrics. Moreover, we propose several cautious, low-risk approaches to using the Isotonic Mechanism and author-provided rankings in peer review processes, including assisting senior area chairs' oversight of area chairs' recommendations, supporting the selection of paper awards, and guiding the recruitment of emergency reviewers. We conclude the paper by addressing the study's limitations and proposing future research directions., Comment: See more details about the experiment at https://openrank.cc/

Gradient-based meta-learning algorithms have gained popularity for their ability to train models on new tasks using limited data. Empirical observations indicate that such algorithms are able to learn a shared representation across tasks, which is regarded as a key factor in their success. However, the in-depth theoretical understanding of the learning dynamics and the origin of the shared representation remains underdeveloped. In this work, we investigate the meta-learning dynamics of the non-linear two-layer neural networks trained on streaming tasks in the teach-student scenario. Through the lens of statistical physics analysis, we characterize the macroscopic behavior of the meta-training processes, the formation of the shared representation, and the generalization ability of the model on new tasks. The analysis also points to the importance of the choice of certain hyper-parameters of the learning algorithms.

The optical properties of graphene (Gr)-covered CsPbBr$_3$ quantum dots (QDs) were investigated using micro-photoluminescence spectroscopy, revealing a remarkable three-orders-of-magnitude enhancement in photoluminescence (PL) intensity compared to bare CsPbBr$_3$ QDs. To elucidate the underlying mechanisms, we combined experimental techniques with density functional theory (DFT) calculations. DFT simulations showed that the graphene layer generates interfacial electrostatic potential barriers when in contact with the CsPbBr$_3$ surface, impeding carrier leakage from perovskite to graphene and enhancing radiative recombination. Additionally, graphene passivates CsPbBr$_3$ surface defect states, suppressing nonradiative recombination of photo-generated carriers. Our study also revealed that graphene becomes n-doped upon contact with CsPbBr$_3$ QDs, activating its plasmon mode. This mode resonantly couples with photo-generated excitons in the perovskite. The momentum mismatch between graphene plasmons and free-space photons is resolved through plasmon scattering at Gr/CsPbBr$_3$ interface corrugations, facilitating the observed super-bright emission. These findings highlight the critical role of graphene as a top contact in dramatically enhancing CsPbBr$_3$ QDs' PL. Our work advances the understanding of graphene-perovskite interfaces and opens new avenues for designing high-efficiency optoelectronic devices. The multifaceted enhancement mechanisms uncovered provide valuable insights for future research in nanophotonics and materials science, potentially leading to breakthroughs in light-emitting technologies., Comment: 33 pages

The international collaboration designing and constructing the Deep Underground Neutrino Experiment (DUNE) at the Long-Baseline Neutrino Facility (LBNF) has developed a two-phase strategy toward the implementation of this leading-edge, large-scale science project. The 2023 report of the US Particle Physics Project Prioritization Panel (P5) reaffirmed this vision and strongly endorsed DUNE Phase I and Phase II, as did the European Strategy for Particle Physics. While the construction of the DUNE Phase I is well underway, this White Paper focuses on DUNE Phase II planning. DUNE Phase-II consists of a third and fourth far detector (FD) module, an upgraded near detector complex, and an enhanced 2.1 MW beam. The fourth FD module is conceived as a "Module of Opportunity", aimed at expanding the physics opportunities, in addition to supporting the core DUNE science program, with more advanced technologies. This document highlights the increased science opportunities offered by the DUNE Phase II near and far detectors, including long-baseline neutrino oscillation physics, neutrino astrophysics, and physics beyond the standard model. It describes the DUNE Phase II near and far detector technologies and detector design concepts that are currently under consideration. A summary of key R&D goals and prototyping phases needed to realize the Phase II detector technical designs is also provided. DUNE's Phase II detectors, along with the increased beam power, will complete the full scope of DUNE, enabling a multi-decadal program of groundbreaking science with neutrinos.

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.

When modifying existing policies in high-risk settings, it is often necessary to ensure with high certainty that the newly proposed policy improves upon a baseline, such as the status quo. In this work, we consider the problem of safe policy improvement, where one only adopts a new policy if it is deemed to be better than the specified baseline with at least pre-specified probability. We focus on threshold policies, a ubiquitous class of policies with applications in economics, healthcare, and digital advertising. Existing methods rely on potentially underpowered safety checks and limit the opportunities for finding safe improvements, so too often they must revert to the baseline to maintain safety. We overcome these issues by leveraging the most powerful safety test in the asymptotic regime and allowing for multiple candidates to be tested for improvement over the baseline. We show that in adversarial settings, our approach controls the rate of adopting a policy worse than the baseline to the pre-specified error level, even in moderate sample sizes. We present CSPI and CSPI-MT, two novel heuristics for selecting cutoff(s) to maximize the policy improvement from baseline. We demonstrate through both synthetic and external datasets that our approaches improve both the detection rates of safe policies and the realized improvement, particularly under stringent safety requirements and low signal-to-noise conditions.