In recent years, there has been increasing interest in automatic facial behavior analysis systems from computing communities such as vision, multimodal interaction, robotics, and affective computing. Building upon the widespread utility of prior open-source facial analysis systems, we introduce OpenFace 3.0, an open-source toolkit capable of facial landmark detection, facial action unit detection, eye-gaze estimation, and facial emotion recognition. OpenFace 3.0 contributes a lightweight unified model for facial analysis, trained with a multi-task architecture across diverse populations, head poses, lighting conditions, video resolutions, and facial analysis tasks. By leveraging the benefits of parameter sharing through a unified model and training paradigm, OpenFACE 3.0 exhibits improvements in prediction performance, inference speed, and memory efficiency over similar toolkits and rivals state-of-theart models. Openface 3.0 can be installed and run with a single line of code and operate in real-time without specialized hardware. OpenFace 3.0 code for training models and running the system is freely available for research purposes and supports contributions from the community.

Real-world formal theorem proving often depends on a wealth of context, including definitions, lemmas, comments, file structure, and other information. We introduce $\texttt{miniCTX}$, which tests a model's ability to prove formal mathematical theorems that depend on new context that is not seen during training. $\texttt{miniCTX}$ contains theorems sourced from real Lean projects and textbooks, each associated with a context that can span tens of thousands of tokens. Models are tasked with proving a theorem given access to code from the theorem's repository, which contains context that is needed for the proof. As a baseline for $\texttt{miniCTX}$, we tested fine-tuning and prompting methods that condition theorem proving on preceding context. Both approaches substantially outperform traditional methods that rely solely on state information. We found that this ability to use context is not captured by previous benchmarks such as $\texttt{miniF2F}$. Alongside $\texttt{miniCTX}$, we offer $\texttt{ntp-toolkit}$ for automatically extracting and annotating theorem proving data, making it easy to add new projects into $\texttt{miniCTX}$ to ensure that contexts are not seen during training. $\texttt{miniCTX}$ offers a challenging and realistic evaluation of neural theorem provers.

Evolving from informatization to intelligence, railway wireless communication system (RWCS) is required to support more high quality service with strained resources. Nevertheless, due to the high mobility of trains, the perfect channel state information (CSI) of RWCS is typically hard to obtain, which may induce severe performance loss. To address these issues, we proposed an active RIS-assisted RSMA RWCS, where the CSI acquired is assumed to be imperfect. In this system, we formulated a weighted sum-rate (WSR) maximization problem to jointly optimize the beamforming, rate allocation vectors and RIS precoding matrix. To overcome the uncertainty of imperfect CSI and non-convexity of this problem, we transform orignial WSR maximization problem into a weighted average sum-rate ones, and we propose a block coordinate descent (BCD) algorithm based on weighted minimum mean square error (WMMSE) , fractional programming (FP) and sample average approximated (SAA) method to solve it. Numerical results show that active RIS-aied rate-splitting architecture has impressive advantages over conventional passive RIS-assisted space division multiple access (SDMA) under imperfect CSI.

To achieve low-latency and high-reliability (LLHR) for the vehicle-to-everything (V2X) services, the Long Term Evolution (LTE)-based solution has been considered as a promising technology. Since the existing LTE networks are based on the orthogonal multiple access (OMA), the limited spectrum resources have not been fully and efficiently utilized, so that severe data congestion and low access efficiency cannot be avoided in dense networks. Non-OMA (NOMA) provides a new solution for 5G V2X services to mitigate traffic congestion and reduce latency. Different from existing works, we propose two relay-assisted NOMA transmission schemes for 5G V2X communications, i.e., half-duplex relay-assisted NOMA (HDR-NOMA) broadcasting/multicasting and full-duplex relay-assisted NOMA (FDR-NOMA) broadcasting/multicasting, and investigate the optimal power allocation problems for them. To improve the quality of service (QoS) for the users with the poor channel conditions and to guarantee the fairness, the power allocation problems are formulated to maximize the minimum achievable rate for all users. Even though both of the formulated problems are neither concave nor convex, it is shown that the considered problems are quasi-concave. Hence, a bisection-based power allocation algorithm is proposed to obtain the optimal solutions to the problems. Numerical results demonstrate that the proposed schemes outperform the scheme with fixed power allocation obviously and achieve a significant performance improvement with respect to a suboptimal fractional transmit power allocation (FTPA) method and the optimized time division multiple access (TDMA) scheme. Besides, compared with HDR-NOMA, better performance can be achieved by FDR-NOMA scheme when the self-interference is sufficiently suppressed.