Article 2200110 by Surita R. Bhatia and co-workers explores the rheology of fluid interfaces with sequence-specific amphiphilic copolymers. Copolymers contain short “microblocks” of 1–6 hydrophobic or hydrophilic repeat units. Results demonstrate that the interfacial rheology is strongly dependent on microblock size. The results provide a new strategy for controlling the dynamics of fluid interfaces using precision polymers.
To investigate schemes of secret key generation from Ultra-wideband (UWB) channel, we study a statistical characterization of UWB outdoor channel for a campus playground scenario based on extensive measurements.Moreover, an efficient secret key generation mechanism exploiting multipath relative delay is developed, and verification of this algorithm is conducted in UWB Line-of-sight (LOS) outdoor channels.For the first time, we compare key-mismatch probability of UWB indoor and outdoor environments.Simulation results demonstrate that the number of multipath proportionally affects key generation rate and key-mismatch probability.In comparison to the conventional method using received signal strength (RSS) as a common random source, our mechanism achieves better performance in terms of common secret bit generation.Simultaneously, security analysis indicates that the proposed scheme can still guarantee security even in the sparse outdoor physical environment free of many reflectors.
The radio frequency (RF) transmitter identification has a wide application prospect in both military and public communications. The traditional RF transmitter identification of technique is mainly based on expert experience, which shows the shortcomings of low recognition accuracy and weak generalization ability. With the fast development in computer vision, deep learning attracts a lot of attention in recent years and is believed to be a promising scheme in RF transmitter identification. In this paper, the RF transmitter identification is studied based on the RF impairment features extracted from the original data. As a typical deep learning scheme, Convolutional Neural Network (CNN) is adopted to train a classifier to distinguish the RF transmitters. The experiment results show that with the proposed classifier, the same-waveform LoRA signals from different transmitters can be identified with very high accuracy.
Multimodal large language models (MLLMs) have shown promising advancements in general visual and language understanding. However, the representation of multimodal information using MLLMs remains largely unexplored. In this work, we introduce a new framework, E5-V, designed to adapt MLLMs for achieving universal multimodal embeddings. Our findings highlight the significant potential of MLLMs in representing multimodal inputs compared to previous approaches. By leveraging MLLMs with prompts, E5-V effectively bridges the modality gap between different types of inputs, demonstrating strong performance in multimodal embeddings even without fine-tuning. We propose a single modality training approach for E5-V, where the model is trained exclusively on text pairs. This method demonstrates significant improvements over traditional multimodal training on image-text pairs, while reducing training costs by approximately 95%. Additionally, this approach eliminates the need for costly multimodal training data collection. Extensive experiments across four types of tasks demonstrate the effectiveness of E5-V. As a universal multimodal model, E5-V not only achieves but often surpasses state-of-the-art performance in each task, despite being trained on a single modality.
In this paper, a novel sparse representation (SR) based method for recognizing targets under different foliage scenarios is presented. Target echo signals of different targets in different scenarios measured by bistatic ultra-wideband (UWB) radar are used for scenarios and targets recognition. Particularly, two overcomplete dictionaries one for scenarios and one for targets are learned from measured real target echo signal waveforms via dictionary learning technique. The SR expresses an input signal as the linear combination of a small number of the learned dictionary elements, which are from the same category as the input signal. The SR of a test target echo signal can be achieved by solving an L 0 -norm minimization problem. For targets recognition under multi-scenario, firstly, we locate the scenario of the targets with the overcomplete dictionary for scenarios via SR. Then, the types of the targets are recognized using the overcomplete dictionary for targets via SR. The effectiveness of the proposed approach is verified by experiments taken in the forest measurement. Experimental results on real measured data show that the proposed method achieves higher recognition accuracy.
Efficient deep learning-based approaches have achieved remarkable performance in single image super-resolution. However, recent studies on efficient super-resolution have mainly focused on reducing the number of parameters and floating-point operations through various network designs. Although these methods can decrease the number of parameters and floating-point operations, they may not necessarily reduce actual running time. To address this issue, we propose a novel multi-stage lightweight network boosting method, which can enable lightweight networks to achieve outstanding performance. Specifically, we leverage enhanced high-resolution output as additional supervision to improve the learning ability of lightweight student networks. Upon convergence of the student network, we further simplify our network structure to a more lightweight level using reparameterization techniques and iterative network pruning. Meanwhile, we adopt an effective lightweight network training strategy that combines multi-anchor distillation and progressive learning, enabling the lightweight network to achieve outstanding performance. Ultimately, our proposed method achieves the fastest inference time among all participants in the NTIRE 2023 efficient super-resolution challenge while maintaining competitive super-resolution performance. Additionally, extensive experiments are conducted to demonstrate the effectiveness of the proposed components. The results show that our approach achieves comparable performance in representative dataset DIV2K, both qualitatively and quantitatively, with faster inference and fewer number of network parameters.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)