Cued Speech (CS) enhances lipreading via hand coding, offering visual phonemic cues that support precise speech perception for the hearing-impaired. The task of CS Video-to-Speech generation (CSV2S) aims to convert CS videos into intelligible speech signals. Most existing research focuses on CS Recognition (CSR), which transcribes video content into text. Consequently, a common solution for CSV2S is to integrate CSR with a text-to-speech (TTS) system. However, this pipeline relies on text as an intermediate medium, which may lead to error propagation and temporal misalignment between speech and CS video dynamics. In contrast, directly generating audio speech from CS video (direct CSV2S) often suffer from the inherent multimodal complexity and the limited availability of CS data. To address these challenges, we propose UniCUE, the first unified framework for CSV2S that directly generates speech from CS videos without relying on intermediate text. The core innovation of UniCUE lies in integrating a understanding task (CSR) that provides fine-grained CS visual-semantic cues to to guide the speech generation. Specifically, UniCUE incorporates a pose-aware visual processor, a semantic alignment pool that enables precise visual–semantic mapping, and a VisioPhonetic adapter to bridge the understanding and generation tasks within a unified architecture. To support this framework, we construct UniCUE-HI, a large-scale Mandarin CS dataset containing 11,282 videos from 14 cuers, including both hearing-impaired and normal-hearing individuals. Extensive experiments conducted on this dataset demonstrate that UniCUE achieves state-of-the-art (SOTA) performance across multiple evaluation metrics.

The escalating security risks in hybrid cloud environments present a critical bottleneck to broader adoption. This study proposes an innovative risk management strategy to maximize provider revenue. By modeling the cloud service architecture as a queueing system, where user requests are “customers” and resources are “servers”, we quantitatively analyze user decision-making under dynamic risk perceptions. Integrating queueing game theory with a Stackelberg framework, we systematically model the customer trade-off between service efficiency and security assurance. Our analysis reveals Nash equilibrium patterns under diverse data breach liability allocations, deriving an optimized provider risk-control model. This strategy guides market behavior through policy optimization to achieve risk-revenue equilibrium. The work advances theoretical foundations for hybrid cloud risk mitigation and provides actionable strategies for cloud providers.

With the rise of the AIGC era, rhythmic music generation has extensive applications, particularly with the surge in motion video creation. However, generating music that is rhythmically synchronized and stylistically aligned with motion video presents significant challenges. Although existing methods have made progress, they still face difficulties in producing high-quality long-term music, particularly when addressing complex rhythmic patterns and maintaining style-consistent musical chords. In this work, we present MotionComposer, a novel retrieval-augmented, easy-to-hard training approach designed to enhance rhythmic music generation. By leveraging the inherent alignment between motion rhythms and music beats, we first tackle the simpler task of beat prediction with BeatNet, which predicts music beats by analyzing motion patterns. To address the complex musical chord generation, we propose ChordNet, a retrieval-augmented network that integrates external data to enrich chord generation. Additionally, to minimize the impact of irrelevant retrievals, we design RAGate, a retrieval adaptive module that selectively filters out low-relevance retrieval references during the retrieval process. Extensive experiments across three scenarios (i.e., dance, figure skating, and floor exercise) demonstrate that our approach significantly enhances video soundtrack generation, achieving new state-of-the-art performance. Our project is available at https://beria-moon.github.io/Soundtrackyour-Motion/.

Speech-to-portrait generation (S2P) plays a crucial role in speech-driven, human-centered creative content generation, aiming to synthesize a speaker’s face portrait with identity consistency from a given speech clip. However, existing S2P methods can typically only preserve attribute consistency, e.g., gender and age, while failing to capture the more important part-appearance consistency due to the coarse speech-face correlation. In this work, we propose Fine-portraitist, a novel retrieval-augmented, easy-to-hard generation framework designed to tackle this problem. Specifically, Fine-portraitist enhances identity consistency in S2P through two key innovations: 1) We first explore the fine-grained speech-face correlation by decomposing the face portrait into speech-related and speech-unrelated parts. Based on this, we propose a two-stage, diffusion-based pipeline to progressively achieve S2P; 2) A retrieval prior is introduced, selected from a retrieval database based on speech feature similarity, providing supplementary external information for more accurate and realistic generation results. Extensive experiments on two datasets, i.e., AVSpeech and VoxCeleb, demonstrate that Fine-portraitist significantly outperforms existing S2P methods.

In many industrial environments, some components fail and cannot be repaired immediately. We propose a novel repair policy for addressing component failure issues in a circular consecutive-$k$-out-of-$n$:F (abbreviated as Cir/Con/k/n:F) system. This repair policy assigns preemptive priority for repair to the component whose breakdown results in system failure (calledemergency repair), while renders anordinary repairto the failed components without causing failure of the system. The ordinary repairs are recorded by the repairman in the order of their failure, which is said that the broken components are stored in “orbit.” When the repairman becomes idle, he makes the orbital search for failed ones according to the first-failed-first-repair discipline, which can be interrupted by an emergency repair. We carry on an extensive investigation on reliability and queueing indices of the considered model. Specifically, we present a Cir/Con/2/6:F system as an example to give sensitivity analysis for the reliability performance. Numerical inversion of Laplace transform–Stehfest method is adopted to obtain approximate solutions for reliability function. Furthermore, the minimization problem of the CBR is addressed by adopting sequential quadratic programming algorithm. This study offers new insights into balancing the expected total repair cost and associated benefits in the Cir/Con/k/n:F system.

This study examines the decision-making interaction between a service provider adopting a congestion-based repair policy and strategic customers in a failure-prone M/M/1 queueing system. The server’s lifetime is exponentially distributed, and a repair starts immediately upon the server’s breakdown. The repair rate is adjustable: the service provider employs a high repair rate (with a high cost) if the number of waiting customers reaches a threshold; otherwise, a low repair rate (with a low cost) is adopted. We model the interaction as a two-stage Stackelberg game: the provider (leader) sets the price, repair threshold, and information policy before customers (followers) decide whether to join. Using backward induction, we characterize the resulting Stackelberg equilibrium. Under fully unobservable and almost unobservable cases, both follow-the-crowd (FTC) and avoid-the-crowd (ATC) behaviors are found to coexist in the customer’s equilibrium joining strategy. Two special models, the classic repair model (when the threshold approaches 0) and the delayed repair model (when the low repair rate approaches 0), are discussed extensively. The classic repair policy maximizes throughput but incurs the highest costs, while delayed repair minimizes costs at the expense of throughput. The proposed congestion-based repair strategy balances these tradeoffs, achieving intermediate throughput and cost levels. Notably, it can increase profits by up to 34.4% compared with classic repair, with its effectiveness amplified under high-cost scenarios. By comparing the unobservable case with the almost unobservable counterpart, we demonstrate that hiding server state information when prices are low and disclosing server information when prices are high can increase profit for the service provider, but at the expense of reducing social welfare.

Breast lesion segmentation in ultrasound images is essential for computer-aided breast-cancer diagnosis. To improve the segmentation performance, most approaches design sophisticated deep-learning models by mining the patterns of foreground lesions and normal backgrounds simultaneously or by unilaterally enhancing foreground lesions via various focal losses. However, the potential of normal backgrounds is underutilized, which could reduce false positives by compacting the feature representation of all normal backgrounds. From a novel viewpoint of bilateral enhancement, we propose a negative-positive cross-attention network to concentrate on normal backgrounds and foreground lesions, respectively. Derived from the complementing opposites of bipolarity in TaiChi, the network is denoted as TaiChiNet, which consists of the negative normal-background and positive foreground-lesion paths. To transmit the information across the two paths, a cross-attention module, a complementary MLP-head, and a complementary loss are built for deep-layer features, shallow-layer features, and mutual-learning supervision, separately. To the best of our knowledge, this is the first work to formulate breast lesion segmentation as a mutual supervision task from the foreground-lesion and normal-background views. Experimental results have demonstrated the effectiveness of TaiChiNet on two breast lesion segmentation datasets with a lightweight architecture. Furthermore, extensive experiments on the thyroid nodule segmentation and retinal optic cup/disc segmentation datasets indicate the application potential of TaiChiNet.

Medical image segmentation is a vital stage in medical image analysis. Numerous deep-learning methods are booming to improve the performance of 2-D medical image segmentation, owing to the fast growth of the convolutional neural network. Generally, the manually defined ground truth is utilized directly to supervise models in the training phase. However, direct supervision of the ground truth often results in ambiguity and distractors as complex challenges appear simultaneously. To alleviate this issue, we propose a gradually recurrent network with curriculum learning, which is supervised by gradual information of the ground truth. The whole model is composed of two independent networks. One is the segmentation network denoted as GREnet, which formulates 2-D medical image segmentation as a temporal task supervised by pixel-level gradual curricula in the training phase. The other is a curriculum-mining network. To a certain degree, the curriculum-mining network provides curricula with an increasing difficulty in the ground truth of the training set by progressively uncovering hard-to-segmentation pixels via a data-driven manner. Given that segmentation is a pixel-level dense-prediction challenge, to the best of our knowledge, this is the first work to function 2-D medical image segmentation as a temporal task with pixel-level curriculum learning. In GREnet, the naive UNet is adopted as the backbone, while ConvLSTM is used to establish the temporal link between gradual curricula. In the curriculum-mining network, UNet++ supplemented by transformer is designed to deliver curricula through the outputs of the modified UNet++ at different layers. Experimental results have demonstrated the effectiveness of GREnet on seven datasets, i.e., three lesion segmentation datasets in dermoscopic images, an optic disc and cup segmentation dataset and a blood vessel segmentation dataset in retinal images, a breast lesion segmentation dataset in ultrasound images, and a lung segmentation dataset in computed tomography (CT).

This paper investigates a hybrid service system with a cloud server and an in-house server. We consider two different scenarios: a hybrid service system with orbit space and a hybrid service system without orbit space. In the hybrid service system with orbit space, customers who fail to enter the cloud server can choose to join the in-house subsystem or to enter an orbit space and retry the cloud server. An admission control mechanism based on queue-length limitation is adopted to adjust whether the cloud service resources are open to customers. When the cloud server cannot be accessed immediately, some customers send their jobs to the in-house subsystem, while others (called opportunists) try to send their jobs to the cloud server again. We obtain the optimal queue-length limitation for a given retrial rate. The service provider and customers are different stakeholders, and their market forces are also different. Therefore, it is more realistic to explore the game relationship between them by using dynamic game theory. We can also explore the joint optimums of the queue-length limitation and the retrial rate in the framework of the Stackelberg game. Finally, by comparing with the hybrid service system without orbit space, we discuss the significance of the existence of orbit space, and gain management insights. It is found that the existence of opportunists may benefit the service provider, although they significantly harm social interests, regardless of whether they are cooperative or non-cooperative; therefore, opportunists are encouraged in some situations. Numerical analysis shows that adding a retrial orbit to a hybrid cloud service system with certain input parameters may even more than triple the service provider’s revenue.

In this paper, a finite-state phase-type semi-Markov channel (FSPHMC) model is proposed for wireless channels with correlated fading. By introducing a phase type distributed sojourn time between channel state transitions, the proposed model generalizes the traditional Finite-State Markov Channel (FSMC) model where the sojourn time in each channel state is assumed to satisfy geometric distribution. Thanks to the flexibility of the phase type distribution, the resultant FSPHMC model is applicable to much wider range of practical fading channels. For facilitating the implementation of FSPHMC model in practical scenarios, a special case, FSPHMC model with negative binomial sojourn time (FSPHMC-NB), is also presented under constraint computational complexity. Compared with the traditional FSMC model, the FSPHMC-NB demonstrates considerable improvement in that it matches the true state duration distribution without a significant increase in complexity. Simulation results are given to validate the flexibility and versatility of the proposed FSPHMC model.