Search

In speaker tracking research, integrating and complementing multi-modal data is a crucial strategy for improving the accuracy and robustness of tracking systems. However, tracking with incomplete modalities remains a challenging issue due to noisy observations caused by occlusion, acoustic noise, and sensor failures. Especially when there is missing data in multiple modalities, the performance of existing multi-modal fusion methods tends to decrease. To this end, we propose a Global-Local Distillation-based Tracker (GLDTracker) for robust audio-visual speaker tracking. GLDTracker is driven by a teacher-student distillation model, enabling the flexible fusion of incomplete information from each modality. The teacher network processes global signals captured by camera and microphone arrays, and the student network handles local information subject to visual occlusion and missing audio channels. By transferring knowledge from teacher to student, the student network can better adapt to complex dynamic scenes with incomplete observations. In the student network, a global feature reconstruction module based on the generative adversarial network is constructed to reconstruct global features from feature embedding with missing local information. Furthermore, a multi-modal multi-level fusion attention is introduced to integrate the incomplete feature and the reconstructed feature, leveraging the complementarity and consistency of audio-visual and global-local features. Experimental results on the AV16.3 dataset demonstrate that the proposed GLDTracker outperforms existing state-of-the-art audio-visual trackers and achieves leading performance on both standard and incomplete modalities datasets, highlighting its superiority and robustness in complex conditions. The code and models will be available., Comment: We request to withdraw our paper from arXiv due to unresolved author disagreements about the data interpretation and study conclusions. To maintain scientific integrity, we believe withdrawing the paper is necessary. We regret any confusion caused

Zero-shot human-object interaction (HOI) detector is capable of generalizing to HOI categories even not encountered during training. Inspired by the impressive zero-shot capabilities offered by CLIP, latest methods strive to leverage CLIP embeddings for improving zero-shot HOI detection. However, these embedding-based methods train the classifier on seen classes only, inevitably resulting in seen-unseen confusion for the model during inference. Besides, we find that using prompt-tuning and adapters further increases the gap between seen and unseen accuracy. To tackle this challenge, we present the first generation-based model using CLIP for zero-shot HOI detection, coined HOIGen. It allows to unlock the potential of CLIP for feature generation instead of feature extraction only. To achieve it, we develop a CLIP-injected feature generator in accordance with the generation of human, object and union features. Then, we extract realistic features of seen samples and mix them with synthetic features together, allowing the model to train seen and unseen classes jointly. To enrich the HOI scores, we construct a generative prototype bank in a pairwise HOI recognition branch, and a multi-knowledge prototype bank in an image-wise HOI recognition branch, respectively. Extensive experiments on HICO-DET benchmark demonstrate our HOIGen achieves superior performance for both seen and unseen classes under various zero-shot settings, compared with other top-performing methods. Code is available at: https://github.com/soberguo/HOIGen, Comment: Accepted by ACM MM 2024

The study of state-to-state spin-exchange collisions in the vicinity of $p$-wave Feshbach resonances offer great opportunities to explore many-body interactions and novel quantum phases. Here, we report the observation of a spin-exchange collision near a $p$-wave Feshbach resonance within a mixture of the lowest and third-lowest hyperfine states of $^6$Li atoms. The spin-exchange interaction is observed over a range of ten gausses and produces a pair of atoms in the second-lowest hyperfine states that are captured by a deep optical dipole trap. We apply a coupled-channel method to calculate the scattering properties of this system. We find that the $p$-wave resonance exhibits a low inelastic collision rate and a broad resonance profile, which is due to the modification by the accompanying spin-exchange collisions. These findings open up new possibilities for the creation of long-lived, strongly interacting $p$-wave Fermi gases.

We theoretically investigate the tunneling transport in a repulsively interacting ultracold Bose-Fermi mixture. A two-terminal model is applied to such a mixture and the supersymmetric-like tunneling current through the junction can be induced by the bias of fermion chemical potential between two reservoirs. The Goldstino, which is the Nambu-Goldstone fermionic mode associated with the spontaneous sypersymmetry breaking and appears as a gapped mode in the presence of the explicit supersymmetry breaking in existing Bose-Fermi mixtures, is found to contribute to the tunneling transport as a supercharge exchanging process. Our study provides a potential way to detect the Goldstino transport in cold atom experiments., Comment: 13 pages, 7 figures

Considering bosonic atoms immersed in a dilute Fermi gas, we theoretically investigate medium-induced bosonic clusters associated with fermion-mediated two- and three-body interactions. Using the variational approach combined with the fermion-mediated interactions, we numerically calculate the binding energies of two- and three-body bosonic clusters in a one-dimensional system. It is found that the bosonic clusters can be formed even with a repulsive boson-boson interaction due to the fermion-mediated interactions. Our results would be relevant for ultracold atomic systems as well as analogue quantum simulations of alpha clusters in neutron-rich matter., Comment: 11 pages, 9 figures

We theoretically investigate in-medium three-body correlations in one-dimensional spinless fermions with antisymmetrized two- and three-body attractive interactions. By investigating the variational problem of three-body states above the Fermi sea, we illuminate the fate of the in-medium three-body cluster states both in the special case with pure attractive three-body interaction as well as in the case with the coexistence of two- and three-body interactions. Our results testify that the fermion-dimer repulsion is canceled by including the three-body interactions, and stable three-body clusters can be formed. We further feature a phase diagram consisting of the $p$-wave Cooper pairing and Cooper tripling phases in a plane of $p$-wave two- and three-body coupling strengths., Comment: 11 pages, 5 figures

Multiple types of Nyquist-like impedance-based criteria are utilized for the small-signal stability analysis of converter-based AC systems. It is usually considered that the determinant-based criterion can determine the overall stability of a system while the eigenvalue-based criterion can give more insights into the mechanism of the instability. This paper specifies such understandings starting with the zero-pole calculation of impedance matrices obtained by state-spaces with the Smith-McMillan form, then clarifying the absolute reliability of determinant-based criterion with the common assumption for impedance-based analysis that each subsystem can stably operate before the interconnection. However, ambiguities do exist for the eigenvalue-based criterion when an anticlockwise encirclement around the origin is observed in the Nyquist plot. To this end, a logarithmic derivative-based criterion to directly identify the system modes using the frequency responses of loop impedances is proposed, which owns a solid theoretical basis of the Schur complement of transfer function matrices. The theoretical analysis is validated using a PSCAD simulation of a grid-connected two-level voltage source converter., Comment: Accepted by CSEE JPES

We theoretically examine a continuity between atomic and molecular Fermi superfluids in a Bose-Fermi mixture near the Feshbach resonance. Considering a two-channel model describing the Feshbach resonance between Fermi and Bose atoms, we have constructed the mean-field framework based on the perturbative expansion of the Feshbach atom-dimer coupling. The resulting effective Hamiltonian exhibits not only the continuity between atom-atom to molecule-molecule Cooper pairings but also becomes equivalent to the two-band-superconductor model with Suhl-Matthias-Walker type pair-exchange coupling. We demonstrate how these atomic and molecular Fermi superfluids coexist within the two-band-like superfluid theory. The pair-exchange coupling and resulting superfluid gaps are found to be strongly enhanced near the Feshbach resonance due to the interplay between the infrared singularity of Bogoliubov phonons and their Landau damping arising from the coupling with fermions. The pair-exchange coupling can be probed via the observation of the intrinsic Josephson effect between atomic and molecular superfluids., Comment: 15 pages, 5 figures

We theoretically investigate in-medium two- and three-body correlations in one-dimensional two-component Fermi gases with coexistent even-parity s-wave and odd-parity p-wave interactions. We find the solutions of the stable in-medium three-body cluster states such as Cooper triple by solving the corresponding in-medium variational equations. We further feature the phase diagram consisting of the s- and p-wave Cooper pairing phase, and Cooper tripling phase, in the plane of s- and p-wave pairing strengths. The Cooper tripling phase dominates over the pairing phases when both s- and p-wave interactions are moderately strong., Comment: 7 pages, 3 figures

We theoretically investigate in-medium two- and three-body correlations in one-dimensional spinless fermions with attractive two-body p-wave interaction. By investigating the variational problem of two- and three-body states above the Fermi sea, we elucidate the fate of the in-medium two- and three-body cluster states. The one-dimensional system with the strong p-wave interaction is found to be stable against the formation of three-body clusters even in the presence of the Fermi sea, in contrast to higher-dimensional systems that suffer the strong three-body loss associated with the trimer formation., Comment: 10 pages, 5 figures

We theoretically study the ground-state properties and the condensations of exciton-like Cooper pairs and biexciton-like Cooper quartets in an electron-hole system. Applying the variational approach associated based on the quartet Bardeen-Cooper-Schrieffer (BCS) model to the four-component fermionic system consisting of spin-1/2 electrons and spin-1/2 holes, we show how Cooper pairs and quartet correlations appear in the equation of state at the thermodynamic limit. The biexciton-like four-body correlations survive even at the high-density regime as a many-body BCS-like state of Cooper quartets. Our results are useful for further understanding of exotic matter in the interdisciplinary context of quantum many-body physics with multiple degrees of freedom., Comment: 10 pages, 4 figures

We investigate the quartet correlations in four-component fermionic systems at the thermodynamic limit within a variational many-body theory. The Bardeen-Cooper-Schrieffer (BCS)-type variational wave function is extended to the systems with the coexistence of pair and quartet correlations at zero temperature. Special attention is paid to the application of the present framework to an alpha-particle condensation in symmetric nuclear matter, where the coexistence of deuteron and alpha condensations is anticipated. We also discuss how physical properties, such as quasiparticle dispersion, can be modified by the pair and quartet correlations and show a hierarchical structure of in-medium cluster formations in infinite nuclear matter. The present results may also contribute to the interdisciplinary understanding of fermionic condensations beyond the BCS paradigm in many-body systems., Comment: 11 pages, 2 figures

Production software oftentimes suffers from the issue of performance inefficiencies caused by inappropriate use of data structures, programming abstractions, and conservative compiler optimizations. It is desirable to avoid unnecessary memory operations. However, existing works often use a whole-program fine-grained monitoring method with incredibly high overhead. To this end, we propose a learning-aided approach to identify unnecessary memory operations intelligently with low overhead. By applying several prevalent graph neural network models to extract program semantics with respect to program structure, execution order and dynamic states, we present a novel, hybrid program embedding approach so that to derive unnecessary memory operations through the embedding. We train our model with tens of thousands of samples acquired from a set of real-world benchmarks. Results show that our model achieves 90% of accuracy and incurs only around a half of time overhead of the state-of-art tool., Comment: 9 pages, 6 figures, 3 tables

Inspired by the reconstituted similarity renormalization group method, the reconstituted Foldy-Wouthuysen (FW) transformation is proposed. Applied to the Dirac equation in the covariant density functional theory, the reconstituted FW transformation shows a fast convergence of the spectrum of the single-particle energy. The single-particle densities and the single-particle scalar densities obtained by this new method are also investigated. In particular, the relativistic corrections to the densities from the picture-change error between the Schr\"odinger and Dirac pictures are discussed in detail. Taking these relativistic corrections into account, both the single-particle densities and the single-particle scalar densities are almost identical to their exact values., Comment: 12 pages, 9 figures

By following the Foldy-Wouthuysen (FW) transformation of the Dirac equation, we work out the exact analytic expressions up to the $1/M^4$ order for the general cases in the covariant density functional theory. These results are further compared with the corresponding ones derived from another novel non-relativistic expansion method, the similarity renormalization group (SRG). Based on that, the origin of the difference between the results obtained by the FW transformation and SRG method is explored., Comment: 14 pages

By following the conventional similarity renormalization group (SRG) expansion of the Dirac equation developed in [J.-Y. Guo, Phys. Rev. C \textbf{85}, 021302 (2012)], we work out the analytic expression of the ${1}/{M^4}$ order and verify the convergence of this method. As a step further, the reconstituted SRG method is proposed by using the re-summation technique. The speed of convergence of the reconstituted SRG becomes much faster than the conventional one, and the single-particle densities with the reconstituted SRG are also almost identical to the exact values., Comment: 10 pages, 3 figures, and 8 tables