Search

Neural Radiance Fields (NeRF) face significant challenges in few-shot scenarios, primarily due to overfitting and long training times for high-fidelity rendering. Existing methods, such as FreeNeRF and SparseNeRF, use frequency regularization or pre-trained priors but struggle with complex scheduling and bias. We introduce FrugalNeRF, a novel few-shot NeRF framework that leverages weight-sharing voxels across multiple scales to efficiently represent scene details. Our key contribution is a cross-scale geometric adaptation scheme that selects pseudo ground truth depth based on reprojection errors across scales. This guides training without relying on externally learned priors, enabling full utilization of the training data. It can also integrate pre-trained priors, enhancing quality without slowing convergence. Experiments on LLFF, DTU, and RealEstate-10K show that FrugalNeRF outperforms other few-shot NeRF methods while significantly reducing training time, making it a practical solution for efficient and accurate 3D scene reconstruction., Comment: Project page: https://linjohnss.github.io/frugalnerf/

We reveal the double pole structures of the manifestly exotic tetraquark state $X_1(2900)$ in the scenario of $P$-wave $\bar{D}^*K^*$ dimeson resonance. We find that the observed enhancement signal associated with $X_1(2900)$ in $B^+ \to D^+D^-K^+$ by LHCb contains two $P$-wave poles denoted as $T_{cs1-}(2900)$ and $T^{\prime}_{cs1-}(2900)$, respectively. After considering the channel couplings among the $\bar{D}K$, $\bar{D}^*K$, $\bar{D}K^*$ and $\bar{D}^*K^*$ and the width of the $K^*$ meson, the masses and widths of the $S$-wave pole $T_{cs0+}(2900)$ and two $P$-wave poles $T_{cs1-}(2900)$ and $T^{\prime}_{cs1-}(2900)$ coincide with those of the $X_0(2900)$ and $X_1(2900)$ remarkably, which provides strong support for identifying $X_0(2900)$ and $X_1(2900)$ as $\bar{D}^{(*)}K^{(*)}$ dimeson states. Furthermore, we extensively calculate all $S$-wave and $P$-wave $\bar{D}^{(*)}K^{(*)}$ systems up to $J=3$ and predict four new isoscalar charmed-strange dimeson-type tetraquark states: an $S$-wave state $T_{cs1+}(2900)$ with quantum number $J^P=1^+$, three $P$-wave states $T_{cs1-}(2760)$ with $J^P=1^-$, $T_{cs0-}(2760)$ with $J^P=0^-$ and $T_{cs2-}(2900)$ with $J^P=2^-$. These near-threshold poles can be searched for at LHCb, Belle II and BESIII., Comment: 12 pages, 5 figures and 4 tables

While Neural Radiance Fields (NeRFs) have demonstrated exceptional quality, their protracted training duration remains a limitation. Generalizable and MVS-based NeRFs, although capable of mitigating training time, often incur tradeoffs in quality. This paper presents a novel approach called BoostMVSNeRFs to enhance the rendering quality of MVS-based NeRFs in large-scale scenes. We first identify limitations in MVS-based NeRF methods, such as restricted viewport coverage and artifacts due to limited input views. Then, we address these limitations by proposing a new method that selects and combines multiple cost volumes during volume rendering. Our method does not require training and can adapt to any MVS-based NeRF methods in a feed-forward fashion to improve rendering quality. Furthermore, our approach is also end-to-end trainable, allowing fine-tuning on specific scenes. We demonstrate the effectiveness of our method through experiments on large-scale datasets, showing significant rendering quality improvements in large-scale scenes and unbounded outdoor scenarios. We release the source code of BoostMVSNeRFs at https://su-terry.github.io/BoostMVSNeRFs/., Comment: SIGGRAPH 2024 Conference Papers. Project page: https://su-terry.github.io/BoostMVSNeRFs/

We present a big-step and small-step operational semantics for Yul -- the intermediate language used by the Solidity compiler to produce EVM bytecode -- in a mathematical notation that is congruous with the literature of programming languages, lends itself to language proofs, and can serve as a precise, widely accessible specification for the language. Our two semantics stay faithful to the original, informal specification of the language but also clarify under-specified cases such as void function calls. Our presentation allows us to prove the equivalence between the two semantics. We also implement the small-step semantics in an interpreter for Yul which avails of optimisations that are provably correct. We have tested the interpreter using tests from the Solidity compiler and our own. We envisage that this work will enable the development of verification and symbolic execution technology directly in Yul, contributing to the Ethereum security ecosystem, as well as aid the development of a provably sound future type system.

This paper introduces a method for zero-shot video restoration using pre-trained image restoration diffusion models. Traditional video restoration methods often need retraining for different settings and struggle with limited generalization across various degradation types and datasets. Our approach uses a hierarchical token merging strategy for keyframes and local frames, combined with a hybrid correspondence mechanism that blends optical flow and feature-based nearest neighbor matching (latent merging). We show that our method not only achieves top performance in zero-shot video restoration but also significantly surpasses trained models in generalization across diverse datasets and extreme degradations (8$\times$ super-resolution and high-standard deviation video denoising). We present evidence through quantitative metrics and visual comparisons on various challenging datasets. Additionally, our technique works with any 2D restoration diffusion model, offering a versatile and powerful tool for video enhancement tasks without extensive retraining. This research leads to more efficient and widely applicable video restoration technologies, supporting advancements in fields that require high-quality video output. See our project page for video results and source code at https://jimmycv07.github.io/DiffIR2VR_web/., Comment: Project page: https://jimmycv07.github.io/DiffIR2VR_web/

Invasive brain-computer interfaces with Electrocorticography (ECoG) have shown promise for high-performance speech decoding in medical applications, but less damaging methods like intracranial stereo-electroencephalography (sEEG) remain underexplored. With rapid advances in representation learning, leveraging abundant recordings to enhance speech decoding is increasingly attractive. However, popular methods often pre-train temporal models based on brain-level tokens, overlooking that brain activities in different regions are highly desynchronized during tasks. Alternatively, they pre-train spatial-temporal models based on channel-level tokens but fail to evaluate them on challenging tasks like speech decoding, which requires intricate processing in specific language-related areas. To address this issue, we collected a well-annotated Chinese word-reading sEEG dataset targeting language-related brain networks from 12 subjects. Using this benchmark, we developed the Du-IN model, which extracts contextual embeddings based on region-level tokens through discrete codex-guided mask modeling. Our model achieves state-of-the-art performance on the 61-word classification task, surpassing all baselines. Model comparisons and ablation studies reveal that our design choices, including (i) temporal modeling based on region-level tokens by utilizing 1D depthwise convolution to fuse channels in the ventral sensorimotor cortex (vSMC) and superior temporal gyrus (STG) and (ii) self-supervision through discrete codex-guided mask modeling, significantly contribute to this performance. Overall, our approach -- inspired by neuroscience findings and capitalizing on region-level representations from specific brain regions -- is suitable for invasive brain modeling and represents a promising neuro-inspired AI approach in brain-computer interfaces.

The $X(3872)$, as the first and the most crucial member in the exotic charmoniumlike $XYZ$ family, has been studied for a long time. However, its dynamical origin, whether stemming from a $D\bar{D}^*$ hadronic molecule or the first excited $P$-wave charmonium $\chi_{c1}(2P)$, remains controversial. In this Letter, we demonstrate that the $X(3872)$ definitely does not result from the mass shift of the higher bare $\chi_{c1}(2P)$ resonance pole in the coupled-channel dynamics involving a short-distance $c\bar{c}$ core and the long-distance $D\bar{D}^*$ channels. Instead, it originates from either the $D\bar{D}^*$ molecular pole or the shadow pole associated with the $P$-wave charmonium, which depends on the concrete coupling mode between the $c\bar{c}$ and $D\bar{D}^*$. In order to further exploit the nature of $X(3872)$, we carefully investigate potential mechanisms that contribute to its pole width, which suggests that the coupled-channel dynamics plays a critical role in causing a noticeable discrepancy between the pole widths of $X(3872)$ and $T_{cc}^+$. Interestingly, we bridge the quantitative connection among the dynamics origin of $X(3872)$, its pole width and the properties of the predicted new resonance. The precise measurement of the pole width of $X(3872)$ and the search for the new charmoniumlike resonance become highly significant and can be anticipated in future LHCb, BESIII and Belle II experiments., Comment: 11 pages, 5 figures

The BESIII Collaboration recently performed a precise measurement of the $e^+e^-\rightarrow D\bar{D}$ Born cross sections, and confirmed the $G(3900)$ structure reported by BaBar and Belle with high significance. We identify the $G(3900)$ as the first P-wave $\DDbar$ molecular resonance. The experimental and theoretical identification of the P-wave dimeson state holds paramount importance in enhancing our comprehension of the non-perturbative QCD and few-body physics. Its existence is firmly established in a unified meson-exchange model which simultaneously depicts the features of the $\chi_{c1}(3872)$, $Z_c(3900)$ and $T_{cc}(3875)$. This scenario can be directly examined in the $e^+e^-\rightarrow D\bar{D}^*/\bar{D}D^*$ cross section by seeing whether a resonance exists at the threshold. The credibility of the investigations is also ensured by the fact that the P-wave interaction dominantly arises from the well-known long-range pion exchange. Additionally, thanks to the centrifugal barrier, it is easier to form resonances in P-wave than in S-wave. We extensively calculate all systems up to P-wave with various quantum numbers and predict a dense population of the $\DDbar$ and $\DD$ states, where the S-wave $\DDbar$ state with $I^G (J^{PC})=0^- (1^{+-})$, P-wave $\DDbar$ state with $I^G(J^{PC})=0^+(0^{-+})$, and P-wave $\DD$ state with $I(J^P)=0(0^-)$ are more likely to be observed in experiments., Comment: 8 pages, 4 figures. PRL(in press)

Stroke is a common disabling neurological condition that affects about one-quarter of the adult population over age 25; more than half of patients still have poor outcomes, such as permanent functional dependence or even death, after the onset of acute stroke. The aim of this study is to investigate the efficacy of diffusion-weighted MRI modalities combining with structured health profile on predicting the functional outcome to facilitate early intervention. A deep fusion learning network is proposed with two-stage training: the first stage focuses on cross-modality representation learning and the second stage on classification. Supervised contrastive learning is exploited to learn discriminative features that separate the two classes of patients from embeddings of individual modalities and from the fused multimodal embedding. The network takes as the input DWI and ADC images, and structured health profile data. The outcome is the prediction of the patient needing long-term care at 3 months after the onset of stroke. Trained and evaluated with a dataset of 3297 patients, our proposed fusion model achieves 0.87, 0.80 and 80.45% for AUC, F1-score and accuracy, respectively, outperforming existing models that consolidate both imaging and structured data in the medical domain. If trained with comprehensive clinical variables, including NIHSS and comorbidities, the gain from images on making accurate prediction is not considered substantial, but significant. However, diffusion-weighted MRI can replace NIHSS to achieve comparable level of accuracy combining with other readily available clinical variables for better generalization., Comment: 12 pages, 5 figures, 5 tables

We propose Pushdown Normal Form (PDNF) Bisimulation to verify contextual equivalence in higher-order functional programming languages with local state. Similar to previous work on Normal Form (NF) bisimulation, PDNF Bisimulation is sound and complete with respect to contextual equivalence. However, unlike traditional NF Bisimulation, PDNF Bisimulation is also decidable for a class of program terms that reach bounded configurations but can potentially have unbounded call stacks and input an unbounded number of unknown functions from their context. Our approach relies on the principle that, in model-checking for reachability, pushdown systems can be simulated by finite-state automata designed to accept their initial/final stack content. We embody this in a stackless Labelled Transition System (LTS), together with an on-the-fly saturation procedure for call stacks, upon which bisimulation is defined. To enhance the effectiveness of our bisimulation, we develop up-to techniques and confirm their soundness for PDNF Bisimulation. We develop a prototype implementation of our technique which is able to verify equivalence in examples from practice and the literature that were out of reach for previous work.

We present the first fully abstract normal form bisimulation for call-by-value PCF (PCF$_{\textsf{v}}$). Our model is based on a labelled transition system (LTS) that combines elements from applicative bisimulation, environmental bisimulation and game semantics. In order to obtain completeness while avoiding the use of semantic quotiening, the LTS constructs traces corresponding to interactions with possible functional contexts. The model gives rise to a sound and complete technique for checking of PCF$_{\textsf{v}}$ program equivalence, which we implement in a bounded bisimulation checking tool. We test our tool on known equivalences from the literature and new examples.

Understanding semantic content from brain activity during sleep represents a major goal in neuroscience. While studies in rodents have shown spontaneous neural reactivation of memories during sleep, capturing the semantic content of human sleep poses a significant challenge due to the absence of well-annotated sleep datasets and the substantial differences in neural patterns between wakefulness and sleep. To address these challenges, we designed a novel cognitive neuroscience experiment and collected a comprehensive, well-annotated electroencephalography (EEG) dataset from 134 subjects during both wakefulness and sleep. Leveraging this benchmark dataset, we developed SI-SD that enhances sleep semantic decoding through the position-wise alignment of neural latent sequence between wakefulness and sleep. In the 15-way classification task, our model achieves 24.12% and 21.39% top-1 accuracy on unseen subjects for NREM 2/3 and REM sleep, respectively, surpassing all other baselines. With additional fine-tuning, decoding performance improves to 30.32% and 31.65%, respectively. Besides, inspired by previous neuroscientific findings, we systematically analyze how the "Slow Oscillation" event impacts decoding performance in NREM 2/3 sleep -- decoding performance on unseen subjects further improves to 40.02%. Together, our findings and methodologies contribute to a promising neuro-AI framework for decoding brain activity during sleep.

This paper introduces a novel voice conversion (VC) model, guided by text instructions such as "articulate slowly with a deep tone" or "speak in a cheerful boyish voice". Unlike traditional methods that rely on reference utterances to determine the attributes of the converted speech, our model adds versatility and specificity to voice conversion. The proposed VC model is a neural codec language model which processes a sequence of discrete codes, resulting in the code sequence of converted speech. It utilizes text instructions as style prompts to modify the prosody and emotional information of the given speech. In contrast to previous approaches, which often rely on employing separate encoders like prosody and content encoders to handle different aspects of the source speech, our model handles various information of speech in an end-to-end manner. Experiments have demonstrated the impressive capabilities of our model in comprehending instructions and delivering reasonable results., Comment: Accepted to ASRU 2023

The three-body threshold effect, the distinctive and intriguing non-perturbative dynamics in the low-energy hadron-hadron scattering, has acquired compelling significance in the wake of the recent observation of the double-charm tetraquark $T_{cc}^+$. This dynamics is characterized by the emergence of singular points and branch cuts within the interaction potential, occurring when the on-shell condition of the mediated particle is satisfied. The presence of these potential singularities indicates that the system is no longer Hermitian and also poses intractable challenges in obtaining exact solutions for dynamical scattering amplitudes. In this work, we develop a complex scaled Lippmann-Schwinger equation as an operation of analytical continuation of the $T$ matrix to resolve this problem. Through a practical application to the $DD^* \to DD^*$ process, we reveal complicated cut structures of the three-body threshold dynamics in the complex plane, primarily stemming from the one-pion exchange. Notably, our methodology succeeds in reproducing the $T_{cc}^+$ structure, in alignment with the quasi-bound pole derived from the complex scaling method within the Schr\"odinger equation framework. More remarkably, after solving the on-shell $T$ matrix on the positive real axis of momentum plane, we find an extra new structure in the $DD^*$ mass spectrum, which arises from a right-hand cut at a physical pion mass and should be observable in Lattice QCD simulations and future high-energy experiments., Comment: 9 pages and 9 figures

We improve the complex scaling method (CSM) to obtain virtual states, which were previously challenging in the conventional CSM. Our approach solves the Schr\"odinger equation in the momentum space as an eigenvalue problem by choosing the flexible contours. It proves to be highly effective in identifying the poles across the different Riemann sheets in the multichannel scatterings. It is more straightforward and efficient than searching for the zeros of the Fredholm determinant of the Lippmann-Schwinger equation using the root-finding algorithms. This advancement significantly extends the capabilities of the CSM in accurately characterizing the resonances and virtual states in quantum systems., Comment: 12 pages, 14 figures, 2 tables. Version accepted by PRD

A partial width formula is proposed using the analytical extension of the wave function in momentum space. The distinction of the Riemann sheets is explained from the perspective of the Schrodinger equation. The analytical form in coordinate space and the partial width are derived subsequently. Then a coupled-channel analysis is performed to investigate the open-charm branching ratios of the $P_c$ states, involving the contact interactions and one-pion-exchange potential with the three-body effects. The low energy constants are fitted using the experimental masses and widths as input. The $P_c(4312)$ is found to decay mainly to $\Lambda_c\bar{D}^*$, while the branching ratios of the $P_c(4440)$ and $P_c(4457)$ in different channels are comparable. Under the reasonable assumption that the off-diagonal contact interactions are small, the $J^P$ quantum numbers of the $P_c(4440)$ and the $P_c(4457)$ prefer $\frac{1}{2}^-$ and $\frac{3}{2}^-$ respectively. Three additional $P_c$ states at 4380 MeV, 4504 MeV and 4516 MeV, together with their branching ratios, are predicted. A deduction of the revised one-pion-exchange potential involving the on-shell three-body intermediate states is performed., Comment: 16 pages, 5 figures

We investigate the $Z_b$, $Z_c$ and $Z_{cs}$ states within the chiral effective field theory framework and the $S$-wave single channel molecule picture. With the complex scaling method, we accurately solve the Schr\"odinger equation in momentum space. Our analysis reveals that the $Z_b(10610)$, $Z_b(10650)$, $Z_c(3900)$ and $Z_c(4020)$ states are the resonances composed of the $S-$wave $(B\bar{B}^{*}+B^{*}\bar{B})/\sqrt{2}$, $B^{*}\bar{B}^*$, $(D\bar{D}^{*}+D^{*}\bar{D})/\sqrt{2}$ and $D^{*}\bar{D}^*$, respectively. Furthermore, although the $Z_{cs}(3985)$ and $Z_{cs}(4000)$ states exhibit a significant difference in width, these two resonances may originate from the same channel, the $S-$wave $(D_{s}\bar{D}^{*}+D_{s}^{*}\bar{D})/\sqrt{2}$. Additionally, we find two resonances in the $S-$wave $D_s^*\bar{D}^*$ channel, corresponding to the $Z_{cs}(4123)$ and $Z_{cs}(4220)$ states that await experimental confirmation., Comment: 10 pages, 5 figures, 4 tables

The utilization of Wi-Fi based human activity recognition has gained considerable interest in recent times, primarily owing to its applications in various domains such as healthcare for monitoring breath and heart rate, security, elderly care. These Wi-Fi-based methods exhibit several advantages over conventional state-of-the-art techniques that rely on cameras and sensors, including lower costs and ease of deployment. However, a significant challenge associated with Wi-Fi-based HAR is the significant decline in performance when the scene or subject changes. To mitigate this issue, it is imperative to train the model using an extensive dataset. In recent studies, the utilization of CNN-based models or sequence-to-sequence models such as LSTM, GRU, or Transformer has become prevalent. While sequence-to-sequence models can be more precise, they are also more computationally intensive and require a larger amount of training data. To tackle these limitations, we propose a novel approach that leverages a temporal convolution network with augmentations and attention, referred to as TCN-AA. Our proposed method is computationally efficient and exhibits improved accuracy even when the data size is increased threefold through our augmentation techniques. Our experiments on a publicly available dataset indicate that our approach outperforms existing state-of-the-art methods, with a final accuracy of 99.42%., Comment: Paper is currently under review at IEEE Access

DAG-based DLTs allow for parallel, asynchronous writing access to a ledger. Consequently, the perception of the most recent blocks may differ considerably between nodes, and the underlying network properties of the P2P layer have a direct impact on the performance of the protocol. Moreover, the stronger inter-dependencies of several core components demand a more complex and complete approach to studying such DLTs. This paper presents an agent-based, open-sourced simulator for large-scale networks that implement the leaderless Tangle 2.0 consensus protocol. Its scope includes modelling the underlying peer-to-peer communication with network topology, package loss, heterogeneous latency, the gossip protocol with reliable broadcast qualities, the underlying DAG-based data structure, and the consensus protocol. The simulator allows us to explore the performance of the protocol in different network environments, as well as different attack scenarios., Comment: IEEE ICBC 2023, short paper

Spontaneous reporting systems (SRS) have been developed to collect adverse event records that contain personal demographics and sensitive information like drug indications and adverse reactions. The release of SRS data may disclose the privacy of the data provider. Unlike other microdata, very few anonymyization methods have been proposed to protect individual privacy while publishing SRS data. MS(k, {\theta}*)-bounding is the first privacy model for SRS data that considers multiple individual records, mutli-valued sensitive attributes, and rare events. PPMS(k, {\theta}*)-bounding then is proposed for solving cross-release attacks caused by the follow-up cases in the periodical SRS releasing scenario. A recent trend of microdata anonymization combines the traditional syntactic model and differential privacy, fusing the advantages of both models to yield a better privacy protection method. This paper proposes the PPMS-DP(k, {\theta}*, {\epsilon}) framework, an enhancement of PPMS(k, {\theta}*)-bounding that embraces differential privacy to improve privacy protection of periodically released SRS data. We propose two anonymization algorithms conforming to the PPMS-DP(k, {\theta}*, {\epsilon}) framework, PPMS-DPnum and PPMS-DPall. Experimental results on the FAERS datasets show that both PPMS-DPnum and PPMS-DPall provide significantly better privacy protection than PPMS-(k, {\theta}*)-bounding without sacrificing data distortion and data utility., Comment: 10 pages, 11 figures

We construct the effective potentials of the $P_c$ and $P_{cs}$ states based on the SU(3)$_{\text{f}}$ symmetry and heavy quark symmetry. Then we perform the coupled-channel analysis of the lowest isospin $P_c$ and $P_{cs}$ systems. The coupled-channel effects play different roles in the $P_c$ and $P_{cs}$ systems. In the $P_c$ systems, this effect gives minor corrections to the masses of the $P_c$ states. In the $P_{cs}$ system, the $\Lambda_c\bar{D}_s-\Xi_c\bar{D}$ coupling will shift the mass of the $P_{cs}(4338)$ close to the $\Xi_c\bar{D}$ threshold. The $\Lambda_c\bar{D}^{(*)}_s-\Xi_c\bar{D}^{(*)}$ coupling will also produce extra $P_{cs}$ states. We discuss the correspondence between the $P_c$ and $P_{cs}$ states. Our results prefer that the SU(3) partners of the observed $P_{c}(4312)$, $P_{c}(4440)$, and $P_{c}(4457)$ in the $P_{cs}$ system have not been found yet., Comment: 10 Pages, 3 figures, and 6 Tables

We investigate the double-charm and hidden-charm hexaquarks as molecules in the framework of the one-boson-exchange potential model. The multichannel coupling and $S-D$ wave mixing are taken into account carefully. We adopt the complex scaling method to investigate the possible quasibound states, whose widths are from the three-body decay channel $\Lambda_c\Lambda_c\pi$ or $\Lambda_c\bar{\Lambda}_c\pi$. For the double-charm system of $I(J^P)=1(1^+)$, we obtain a quasibound state, whose width is 0.50 MeV if the binding energy is -14.27 MeV. And the $S$-wave $\Lambda_c\Sigma_c$ and $\Lambda_c\Sigma_c^*$ components give the dominant contributions. For the $1(0^+)$ double-charm hexaquark system, we do not find any pole. We find more poles in the hidden-charm hexaquark system. We obtain one pole as a quasibound state in the $I^G(J^{PC})=1^+(0^{--})$ system, which only has one channel $(\Lambda_c\bar{\Sigma}_c+\Sigma_c\bar{\Lambda}_c)/\sqrt{2}$. Its width is 1.72 MeV with a binding energy of -5.37 MeV. But, we do not find any pole for the scalar $1^-(0^{-+})$ system. For the vector $1^-(1^{-+})$ system, we find a quasibound state. Its energies, widths and constituents are very similar to those of the $1(1^+)$ double-charm case. In the vector $1^+(1^{--})$ system, we get two poles -- a quasibound state and a resonance. The quasibound state has a width of 0.6 MeV with a binding energy of -15.37 MeV. For the resonance, its width is 2.72 MeV with an energy of 63.55 MeV relative to the $\Lambda_c\bar{\Sigma}_c$ threshold. And its partial width from the two-body decay channel $(\Lambda_c\bar{\Sigma}_c-\Sigma_c\bar{\Lambda}_c)/\sqrt{2}$ is apparently larger than the partial width from the three-body decay channel $\Lambda_c\bar{\Lambda}_c\pi$.