The detection of patch presence test plays a crucial role in preventing 1-day vulnerability exploitation. Current approaches however face significant challenges in cross-architecture scenarios due to compiler optimization variations and code obfuscation techniques, manifesting in elevated false positive rates and limited generalization. This paper presents MR-Patch, a Multimodal Retrieval-Augmented Patch Verification system that synergizes static analysis with large language model capabilities. The framework employs context-sensitive basic block signature mapping to isolate security-relevant code regions, constructing enriched semantic representations through the joint modeling of control flow graphs, data flow graphs, and instruction-level semantics. A heterogeneous-aware embedding model dynamically fuses these tri-modal features using an adaptive weight allocation mechanism that resists compilation induced distortions. The system’s innovation lies in its hierarchical verification architecture: initial similarity matching via multimodal retrieval is augmented by a locally deployed expert LLM performing contextual semantic validation.Experimental results demonstrate that MR-Patch attains an accuracy of 89.1 % on a test set comprising 6,980 cross-version functions, thereby validating the efficacy of large language models in patch detection.

Respiration rate is an important healthcare indicator, and it has become a popular research topic in remote healthcare applications with Internet of Things. Existing respiration monitoring systems have limitations in terms of convenience, comfort, and privacy, etc. This paper presents a contactless and real-time respiration monitoring system, the so-called Wi-Breath, based on off-the-shelf WiFi devices. The system monitors respiration with both the amplitude and phase difference of the WiFi channel state information (CSI), which is sensitive to human body micro movement. The phase information of the CSI signal is considered and both the amplitude and phase difference are used. For better respiration detection accuracy, a signal selection method is proposed to select an appropriate signal from the amplitude and phase difference based on a support vector machine (SVM) algorithm. Experimental results demonstrate that the Wi-Breath achieves an accuracy of 91.2% for respiration detection, and has a 17.0% reduction in average error in comparison with state-of-the-art counterparts.

Most existing attention-based methods on image captioning focus on the current word and visual information in one time step and generate the next word, without considering the visual and linguistic coherence. We propose Look Back (LB) method to embed visual information from the past and Predict Forward (PF) approach to look into future. LB method introduces attention value from the previous time step into the current attention generation to suit visual coherence of human. PF model predicts the next two words in one time step and jointly employs their probabilities for inference. Then the two approaches are combined together as LBPF to further integrate visual information from the past and linguistic information in the future to improve image captioning performance. All the three methods are applied on a classic base decoder, and show remarkable improvements on MSCOCO dataset with small increments on parameter counts. Our LBPF model achieves BLEU-4 / CIDEr / SPICE scores of 37.4 / 116.4 / 21.2 with cross-entropy loss and 38.3 / 127.6 / 22.0 with CIDEr optimization. Our three proposed methods can be easily applied on most attention-based encoder-decoder models for image captioning.

Neural networks with both high accuracy and small network size are urgently required for mobile phone applications. However, previous network search methods do not take network size into account. In this paper, we use the reinforcement learning method to search for networks offline, with both high accuracy and small network size. Gaussian policy is used to explore the number of convolutional channels in a finer manner. Parameter reward is included in our reward function to punish large networks. We also use binary networks to further reduce network size. Without skip connections or branches, the network generated by our method is competitive with other methods on Cifar-10. Our network is much smaller than networks generated by other network search methods. Besides, our accuracy is higher than original binary network reported in BinaryConnect and is competitive with other real-valued networks.

Standard convolutional neural networks assemble multiple convolutional layers to extract high-level features. Recent efforts keep designing deeper and wider architectures. Even with skip connections applied to combine different layers, the useful low-level features are not effectively utilized. Some deep layer aggregation methods have been proposed to aggregate features of all levels, using simple linear combination or complex non-linear transformation. In this paper, we treat convolutional features as a sequence, and propose our Recurrent Aggregation of Convolutional Neural Network (CNN-RA). Our aggregation method splits a standard CNN into blocks and maps their feature matrices to a sequence of vectors of the same length. LSTM is employed to connect to the sequence and better fuse features across blocks. Our proposed CNN-RA can be directly appended to any standard CNN without any modifications. Experiments show remarkable improvements of CNN-RA over the original architectures across datasets.