arXiv subjects

Wei Song

Publications and source records attributed to Wei Song.

At least 145 records · Page 8Linked to original sources

arXiv2023

Robust Calibrate Proxy Loss for Deep Metric Learning

The mainstream researche in deep metric learning can be divided into two genres: proxy-based and pair-based methods. Proxy-based methods have attracted extensive attention due to the lower training complexity and fast network convergence. However, these methods have limitations as the poxy optimization is done by network, which makes it challenging for the proxy to accurately represent the feature distrubtion of the real class of data. In this paper, we propose a Calibrate Proxy (CP) structure, which uses the real sample information to improve the similarity calculation in proxy-based loss and introduces a calibration loss to constraint the proxy optimization towards the center of the class features. At the same time, we set a small number of proxies for each class to alleviate the impact of intra-class differences on retrieval performance. The effectiveness of our method is evaluated by extensive experiments on three public datasets and multiple synthetic label-noise datasets. The results show that our approach can effectively improve the performance of commonly used proxy-based losses on both regular and noisy datasets.

cs.IR↗

arXiv2023

Maximal co-occurrence nonoverlapping sequential rule mining

The aim of sequential pattern mining (SPM) is to discover potentially useful information from a given se-quence. Although various SPM methods have been investigated, most of these focus on mining all of the patterns. However, users sometimes want to mine patterns with the same specific prefix pattern, called co-occurrence pattern. Since sequential rule mining can make better use of the results of SPM, and obtain better recommendation performance, this paper addresses the issue of maximal co-occurrence nonoverlapping sequential rule (MCoR) mining and proposes the MCoR-Miner algo-rithm. To improve the efficiency of support calculation, MCoR-Miner employs depth-first search and backtracking strategies equipped with an indexing mechanism to avoid the use of sequential searching. To obviate useless support calculations for some sequences, MCoR-Miner adopts a filtering strategy to prune the sequences without the prefix pattern. To reduce the number of candidate patterns, MCoR-Miner applies the frequent item and binomial enumeration tree strategies. To avoid searching for the maximal rules through brute force, MCoR-Miner uses a screening strategy. To validate the per-formance of MCoR-Miner, eleven competitive algorithms were conducted on eight sequences. Our experimental results showed that MCoR-Miner outperformed other competitive algorithms, and yielded better recommendation performance than frequent co-occurrence pattern mining. All algorithms and datasets can be downloaded from https://github.com/wuc567/Pattern-Mining/tree/master/MCoR-Miner.

cs.DB↗

arXiv2022

Multi-Speaker Multi-Style Speech Synthesis with Timbre and Style Disentanglement

Disentanglement of a speaker's timbre and style is very important for style transfer in multi-speaker multi-style text-to-speech (TTS) scenarios. With the disentanglement of timbres and styles, TTS systems could synthesize expressive speech for a given speaker with any style which has been seen in the training corpus. However, there are still some shortcomings with the current research on timbre and style disentanglement. The current method either requires single-speaker multi-style recordings, which are difficult and expensive to collect, or uses a complex network and complicated training method, which is difficult to reproduce and control the style transfer behavior. To improve the disentanglement effectiveness of timbres and styles, and to remove the reliance on single-speaker multi-style corpus, a simple but effective timbre and style disentanglement method is proposed in this paper. The FastSpeech2 network is employed as the backbone network, with explicit duration, pitch, and energy trajectory to represent the style. Each speaker's data is considered as a separate and isolated style, then a speaker embedding and a style embedding are added to the FastSpeech2 network to learn disentangled representations. Utterance level pitch and energy normalization are utilized to improve the decoupling effect. Experimental results demonstrate that the proposed model could synthesize speech with any style seen during training with high style similarity while maintaining very high speaker similarity.

cs.SD↗

arXiv2022

Remapped Cache Layout: Thwarting Cache-Based Side-Channel Attacks with a Hardware Defense

As cache-based side-channel attacks become serious security problems, various defenses have been proposed and deployed in both software and hardware. Consequently, cache-based side-channel attacks on processes co-residing on the same core are becoming extremely difficult. Most of recent attacks then shift their focus to the last-level cache (LLC). Although cache partitioning is currently the most promising defense against the attacks abusing LLC, it is ineffective in thwarting the side-channel attacks that automatically create eviction sets or bypass the user address space layout randomization. In fact, these attacks are largely undefended in current computer systems. We propose Remapped Cache Layout (\textsf{RCL}) -- a pure hardware defense against a broad range of conflict-based side-channel attacks. \textsf{RCL} obfuscates the mapping from address to cache sets; therefore, an attacker cannot accurately infer the location of her data in caches or using a cache set to infer her victim's data. To our best knowledge, it is the first defense to thwart the aforementioned largely undefended side-channel attacks . \textsf{RCL} has been implemented in a superscalar processor and detailed evaluation results show that \textsf{RCL} incurs only small costs in area, frequency and execution time.

cs.CR↗

arXiv2022

Singing Voice Synthesis with Vibrato Modeling and Latent Energy Representation

This paper proposes an expressive singing voice synthesis system by introducing explicit vibrato modeling and latent energy representation. Vibrato is essential to the naturalness of synthesized sound, due to the inherent characteristics of human singing. Hence, a deep learning-based vibrato model is introduced in this paper to control the vibrato's likeliness, rate, depth and phase in singing, where the vibrato likeliness represents the existence probability of vibrato and it would help improve the singing voice's naturalness. Actually, there is no annotated label about vibrato likeliness in existing singing corpus. We adopt a novel vibrato likeliness labeling method to label the vibrato likeliness automatically. Meanwhile, the power spectrogram of audio contains rich information that can improve the expressiveness of singing. An autoencoder-based latent energy bottleneck feature is proposed for expressive singing voice synthesis. Experimental results on the open dataset NUS48E show that both the vibrato modeling and the latent energy representation could significantly improve the expressiveness of singing voice. The audio samples are shown in the demo website.

cs.SD↗

arXiv2022

End-To-End Audiovisual Feature Fusion for Active Speaker Detection

Active speaker detection plays a vital role in human-machine interaction. Recently, a few end-to-end audiovisual frameworks emerged. However, these models' inference time was not explored and are not applicable for real-time applications due to their complexity and large input size. In addition, they explored a similar feature extraction strategy that employs the ConvNet on audio and visual inputs. This work presents a novel two-stream end-to-end framework fusing features extracted from images via VGG-M with raw Mel Frequency Cepstrum Coefficients features extracted from the audio waveform. The network has two BiGRU layers attached to each stream to handle each stream's temporal dynamic before fusion. After fusion, one BiGRU layer is attached to model the joint temporal dynamics. The experiment result on the AVA-ActiveSpeaker dataset indicates that our new feature extraction strategy shows more robustness to noisy signals and better inference time than models that employed ConvNet on both modalities. The proposed model predicts within 44.41 ms, which is fast enough for real-time applications. Our best-performing model attained 88.929% accuracy, nearly the same detection result as state-of-the-art -work.

cs.SD↗

arXiv2022

A BMS-invariant free scalar model

The BMS (Bondi-van der Burg-Metzner-Sachs) symmetry arises as the asymptotic symmetry of flat spacetime at null infinity. In particular, the BMS algebra for three dimensional flat spacetime (BMS$_3$) is generated by the super-rotation generators which form a Virasoro sub-algebra with central charge $c_L$, together with mutually-commuting super-translation generators. The super-rotation and super-translation generators have non-trivial commutation relations with another central charge $c_M$. In this paper, we study a free scalar theory in two dimensions exhibiting BMS$_3$ symmetry, which can also be understood as the ultra-relativistic limit of a free scalar CFT$_2$. Upon canonical quantization on the highest weight vacuum, the central charges are found to be $c_L=2$ and $c_M=0$. Because of the vanishing central charge $c_M=0$, the theory features novel properties: there exist primary states which form a multiplet, and the Hilbert space can be organized by an enlarged version of BMS modules dubbed the staggered modules. We further calculate correlation functions and the torus partition function, the later of which is also shown explicitly to be modular invariant.

hep-th↗

arXiv2022

TGRMPT: A Head-Shoulder Aided Multi-Person Tracker and a New Large-Scale Dataset for Tour-Guide Robot

A service robot serving safely and politely needs to track the surrounding people robustly, especially for Tour-Guide Robot (TGR). However, existing multi-object tracking (MOT) or multi-person tracking (MPT) methods are not applicable to TGR for the following reasons: 1. lacking relevant large-scale datasets; 2. lacking applicable metrics to evaluate trackers. In this work, we target the visual perceptual tasks for TGR and present the TGRDB dataset, a novel large-scale multi-person tracking dataset containing roughly 5.6 hours of annotated videos and over 450 long-term trajectories. Besides, we propose a more applicable metric to evaluate trackers using our dataset. As part of our work, we present TGRMPT, a novel MPT system that incorporates information from head shoulder and whole body, and achieves state-of-the-art performance. We have released our codes and dataset in https://github.com/wenwenzju/TGRMPT.

cs.CV↗

arXiv2022

Monogamy of quantum entanglement

Unlike classical correlation, quantum entanglement cannot be freely shared among many parties. This restricted shareability of entanglement among multi-party systems is known as monogamy of entanglement, which is one of the most fundamental properties of entanglement. Here, we summarize recent theoretical progress in the field of monogamy of entanglement. We firstly review the standard CKW-type monogamy inequalities in terms of various entanglement measures. In particular, the squashed entanglement and one-way distillable entanglement are monogamous for arbitrary dimensional systems. We then introduce some generalized version of monogamy inequalities which extend and sharpen the traditional ones. We also consider the dual polygamy inequalities for multi-party systems. Moreover, we present two new definitions to define monogamy of entanglement. Finally, some challenges and future directions for monogamy of entanglement are highlighted.

quant-ph↗

arXiv2022

A Systematic Review on Affective Computing: Emotion Models, Databases, and Recent Advances

Affective computing plays a key role in human-computer interactions, entertainment, teaching, safe driving, and multimedia integration. Major breakthroughs have been made recently in the areas of affective computing (i.e., emotion recognition and sentiment analysis). Affective computing is realized based on unimodal or multimodal data, primarily consisting of physical information (e.g., textual, audio, and visual data) and physiological signals (e.g., EEG and ECG signals). Physical-based affect recognition caters to more researchers due to multiple public databases. However, it is hard to reveal one's inner emotion hidden purposely from facial expressions, audio tones, body gestures, etc. Physiological signals can generate more precise and reliable emotional results; yet, the difficulty in acquiring physiological signals also hinders their practical application. Thus, the fusion of physical information and physiological signals can provide useful features of emotional states and lead to higher accuracy. Instead of focusing on one specific field of affective analysis, we systematically review recent advances in the affective computing, and taxonomize unimodal affect recognition as well as multimodal affective analysis. Firstly, we introduce two typical emotion models followed by commonly used databases for affective computing. Next, we survey and taxonomize state-of-the-art unimodal affect recognition and multimodal affective analysis in terms of their detailed architectures and performances. Finally, we discuss some important aspects on affective computing and their applications and conclude this review with an indication of the most promising future directions, such as the establishment of baseline dataset, fusion strategies for multimodal affective analysis, and unsupervised learning models.

cs.MM↗

arXiv2021

A Comprehensive and Cross-Platform Test Suite for Memory Safety -- Towards an Open Framework for Testing Processor Hardware Supported Security Extensions

Memory safety remains a critical and widely violated property in reality. Numerous defense techniques have been proposed and developed but most of them are not applied or enabled by default in production-ready environment due to their substantial running cost. The situation might change in the near future because the hardware supported defenses against these attacks are finally beginning to be adopted by commercial processors, operating systems and compilers. We then face a question as there is currently no suitable test suite to measure the memory safety extensions supported on different processors. In fact, the issue is not constrained only for memory safety but all aspect of processor security. All of the existing test suites related to processor security lack some of the key properties, such as comprehensiveness, distinguishability and portability. As an initial step, we propose an expandable test framework for measuring the processor security and open source a memory safety test suite utilizing this framework. The framework is deliberately designed to be flexible so it can be gradually extended to all types of hardware supported security extensions in processors. The initial test suite for memory safety currently contains 160 test cases covering spatial and temporal safety of memory, memory access control, pointer integrity and control-flow integrity. Each type of vulnerabilities and their related defenses have been individually evaluated by one or more test cases. The test suite has been ported to three different instruction set architectures (ISAs) and experimented on six different platforms. We have also utilized the test suite to explore the security benefits of applying different sets of compiler flags available on the latest GNU GCC and LLVM compilers.

cs.CR↗

arXiv2021

Stable and scalable multistage terahertz-driven particle accelerator

Particle accelerators that use electromagnetic fields to increase a charged particle's energy have greatly advanced the development of science and industry since invention. However, the enormous cost and size of conventional radio-frequency accelerators have limited their accessibility. Here we demonstrate a mini-accelerator powered by terahertz pulses with wavelengths 100 times shorter than radio-frequency pulses. By injecting a short relativistic electron bunch to a 30-mm-long dielectric-lined waveguide and tuning the frequency of a 20-period terahertz pulse to the phase-velocity-matched value, precise and sustained acceleration for nearly 100% of the electrons is achieved with the beam energy spread essentially unchanged. Furthermore, by accurately controlling the phase of two terahertz pulses, the beam is stably accelerated successively in two dielectric waveguides with close to 100% charge coupling efficiency. Our results demonstrate stable and scalable beam acceleration in a multistage mini-accelerator and pave the way for functioning terahertz-driven high-energy accelerators.

physics.acc-ph↗

arXiv2021

Gravitational perturbations from NHEK to Kerr

We revisit the spectrum of linear axisymmetric gravitational perturbations of the (near-)extreme Kerr black hole. Our aim is to characterise those perturbations that are responsible for the deviations away from extremality, and to contrast them with the linearized perturbations treated in the Newman-Penrose formalism. For the near horizon region of the (near-)extreme Kerr solution, i.e. the (near-)NHEK background, we provide a complete characterisation of axisymmetric modes. This involves an infinite tower of propagating modes together with the much subtler low-lying mode sectors that contain the deformations driving the black hole away from extremality. Our analysis includes their effects on the line element, their contributions to Iyer-Wald charges around the NHEK geometry, and how to reconstitute them as gravitational perturbations on Kerr. We present in detail how regularity conditions along the angular variables modify the dynamical properties of the low-lying sector, and in particular their role in the new developments of nearly-AdS$_2$ holography.

hep-th↗

arXiv2021

AdS$_3$ gravity and the complex SYK models

We provide a non conformal generalization of the Compère-Song-Strominger (CSS) boundary conditions for AdS$_3$ gravity that breaks the $\widehat u(1)$ Kac-Moody-Virasoro symmetry to two $u(1)$s. The holographic dual specified by the new boundary conditions can be understood as an irrelevant deformation of a warped conformal field theory (WCFT). Upon consistent reduction to two dimensions, AdS$_3$ gravity results in a deformed Jackiw-Teitelboim dilaton gravity model coupled to a Maxwell field. We show that near extremality the boundary conditions inherited from generalized CSS boundary conditions in three dimensions give rise to an effective action exhibiting the same symmetry breaking pattern as the complex Sachdev-Ye-Kitaev models. Besides the Schwarzian term reflecting the breaking of conformal symmetry, the effective action contains an additional term that captures the breaking of the $\widehat u(1)$ Kac-Moody symmetry.

hep-th↗

arXiv2021

MAB-Malware: A Reinforcement Learning Framework for Attacking Static Malware Classifiers

Modern commercial antivirus systems increasingly rely on machine learning to keep up with the rampant inflation of new malware. However, it is well-known that machine learning models are vulnerable to adversarial examples (AEs). Previous works have shown that ML malware classifiers are fragile to the white-box adversarial attacks. However, ML models used in commercial antivirus products are usually not available to attackers and only return hard classification labels. Therefore, it is more practical to evaluate the robustness of ML models and real-world AVs in a pure black-box manner. We propose a black-box Reinforcement Learning (RL) based framework to generate AEs for PE malware classifiers and AV engines. It regards the adversarial attack problem as a multi-armed bandit problem, which finds an optimal balance between exploiting the successful patterns and exploring more varieties. Compared to other frameworks, our improvements lie in three points. 1) Limiting the exploration space by modeling the generation process as a stateless process to avoid combination explosions. 2) Due to the critical role of payload in AE generation, we design to reuse the successful payload in modeling. 3) Minimizing the changes on AE samples to correctly assign the rewards in RL learning. It also helps identify the root cause of evasions. As a result, our framework has much higher black-box evasion rates than other off-the-shelf frameworks. Results show it has over 74\%--97\% evasion rate for two state-of-the-art ML detectors and over 32\%--48\% evasion rate for commercial AVs in a pure black-box setting. We also demonstrate that the transferability of adversarial attacks among ML-based classifiers is higher than the attack transferability between purely ML-based and commercial AVs.

cs.CR↗

arXiv2021

Gated Transformer Networks for Multivariate Time Series Classification

Deep learning model (primarily convolutional networks and LSTM) for time series classification has been studied broadly by the community with the wide applications in different domains like healthcare, finance, industrial engineering and IoT. Meanwhile, Transformer Networks recently achieved frontier performance on various natural language processing and computer vision tasks. In this work, we explored a simple extension of the current Transformer Networks with gating, named Gated Transformer Networks (GTN) for the multivariate time series classification problem. With the gating that merges two towers of Transformer which model the channel-wise and step-wise correlations respectively, we show how GTN is naturally and effectively suitable for the multivariate time series classification task. We conduct comprehensive experiments on thirteen dataset with full ablation study. Our results show that GTN is able to achieve competing results with current state-of-the-art deep learning models. We also explored the attention map for the natural interpretability of GTN on time series modeling. Our preliminary results provide a strong baseline for the Transformer Networks on multivariate time series classification task and grounds the foundation for future research.

cs.LG↗

arXiv2021

Criminal Networks Analysis in Missing Data scenarios through Graph Distances

Data collected in criminal investigations may suffer from: (i) incompleteness, due to the covert nature of criminal organisations; (ii) incorrectness, caused by either unintentional data collection errors and intentional deception by criminals; (iii) inconsistency, when the same information is collected into law enforcement databases multiple times, or in different formats. In this paper we analyse nine real criminal networks of different nature (i.e., Mafia networks, criminal street gangs and terrorist organizations) in order to quantify the impact of incomplete data and to determine which network type is most affected by it. The networks are firstly pruned following two specific methods: (i) random edges removal, simulating the scenario in which the Law Enforcement Agencies (LEAs) fail to intercept some calls, or to spot sporadic meetings among suspects; (ii) nodes removal, that catches the hypothesis in which some suspects cannot be intercepted or investigated. Finally we compute spectral (i.e., Adjacency, Laplacian and Normalised Laplacian Spectral Distances) and matrix (i.e., Root Euclidean Distance) distances between the complete and pruned networks, which we compare using statistical analysis. Our investigation identified two main features: first, the overall understanding of the criminal networks remains high even with incomplete data on criminal interactions (i.e., 10% removed edges); second, removing even a small fraction of suspects not investigated (i.e., 2% removed nodes) may lead to significant misinterpretation of the overall network.

cs.SI↗

arXiv2020

Swing surfaces and holographic entanglement beyond AdS/CFT

We propose a holographic entanglement entropy prescription for general states and regions in two models of holography beyond AdS/CFT known as flat$_3$/BMSFT and (W)AdS$_3$/WCFT. Flat$_3$/BMSFT is a candidate of holography for asymptotically flat three-dimensional spacetimes, while (W)AdS$_3$/WCFT is relevant in the study of black holes in the real world. In particular, the boundary theories are examples of quantum field theories that feature an infinite dimensional symmetry group but break Lorentz invariance. Our holographic entanglement entropy proposal is given by the area of a swing surface that consists of ropes, which are null geodesics emanating from the entangling surface at the boundary, and a bench, which is a spacelike geodesic connecting the ropes. The proposal is supported by an extension of the Lewkowycz-Maldacena argument, reproduces previous results based on the Rindler method, and satisfies the first law of entanglement entropy.

hep-th↗