Enhanced spatiotemporal skeleton modeling: integrating part-joint attention with dynamic graph convolution. [PDF]
Qin Y, Liu S, Yuan C.
europepmc +1 more source
Optical Neuromorphic Technology Catalyzes the Next‐Generation Mobile Communication Technology
Advancements in optical computing could revolutionize next‐generation wireless communication. The review highlights the role of photonic integrated circuits in implementing neural network operations and discusses their benefits, such as high computational density.
Xiaoxiong Song +6 more
wiley +1 more source
Generative Sign-Description Prompts with Multi-Positive Contrastive Learning for Sign Language Recognition. [PDF]
Liang S +6 more
europepmc +1 more source
Semantics of ‘State vs. Action’ in the Content of Perception Verbs [PDF]
Olga Yuryevna Avdevnina
openalex +1 more source
Skin‐Like Airflow Odometry for Micro Aerial Vehicles Based on Distributed Thermal Anemometers
A novel bioinspired skin‐like airflow odometry system for micro aerial vehicles employs distributed thermal anemometers to enable motion estimation in GNSS‐denied environments. Computational fluid dynamics simulation validates sensor deployment feasibility, while a transformer‐enhanced gated recurrent unit network extracts high‐precision airflow ...
Weicheng Di +7 more
wiley +1 more source
The grammatical status of the Polish verbal noun of action and the semantics of delimitativity [PDF]
Елена Эдуардовна Пчелинцева
openalex +1 more source
Anchoring Morphological Representations Unlocks Latent Proprioception in Soft Robots
The authors present ProSoRo, a proprioceptive soft robot that anchors morphological states to a single internal reference frame and employs a multimodal variational autoencoder for crossmodal inference, revealing latent proprioception that enables accurate force and shape estimation, interpretable morphing primitives, robust sim‐to‐real transfer, and ...
Xudong Han +4 more
wiley +1 more source
GA: MCFFS method architecture. Pathological images are processed by the multi‐level feature information fusion (MLFF) method and the multi‐scale feature map fusion strategy to achieve the recognition result of cell nuclei. The MLFF module contains convolution blocks, spatial self‐attention, and atrous spatial pyramid pooling.
Yingqing Lu +3 more
wiley +1 more source
From Speech Semantics to Brain Activity-Timescales Are Key in Their Information Transfer. [PDF]
Kumar S +5 more
europepmc +1 more source

