The rapid rise of real-time communication and large language models has significantly increased the importance of speech compression. Deep learning-based neural speech codecs have outperformed traditional signal-level speech codecs in terms of rate-distortion (RD) performance. Typically, these neural codecs employ an encoder-quantizer-decoder architecture, where audio is first converted into latent code feature representations and then into discrete tokens. However, this architecture exhibits insufficient RD performance due to two main drawbacks: (1) the inadequate performance of the quantizer, challenging training processes, and issues such as codebook collapse; (2) the limited representational capacity of the encoder and decoder, making it difficult to meet feature representation requirements across various bitrates. In this paper, we propose a rate-aware learned speech compression scheme that replaces the quantizer with an advanced channel-wise entropy model to improve RD performance, simplify training, and avoid codebook collapse. We employ multi-scale convolution and linear attention mixture blocks to enhance the representational capacity and flexibility of the encoder and decoder. Experimental results demonstrate that the proposed method achieves state-of-the-art RD performance, obtaining 53.51% BD-Rate bitrate saving in average, and achieves 0.26 BD-VisQol and 0.44 BD-PESQ gains.

Affected by the COVID-19 pandemic, teleworking is becoming more popular, with the exposed attack surface of the internal network expanding. Once outsiders personate accounts or insiders conduct illegal operations, the data security in teleworking with traditional border protection will be broken. Therefore, it is necessary to implement fine-grained and dynamic access control to protect data from malicious access. Attribute-based access control (ABAC) is ideal, where authorization is performed through attributes and rules. On this basis, risk assessment, context awareness, and machine learning are supplemented for dynamic access control. However, these methods have their limitations due to the requirement of sufficient prior knowledge and massive label-classified data. Moreover, it is challenging to obtain the samples of attack behaviors, and the attack behaviors may change frequently to evade detection. In contrast, the normal behaviors are relatively stable except for the update of network services. We propose a dynamic access control model, ABAC-IntroVAE, to address the above issues. ABAC-IntroVAE judges users' requests through rule matching and behavior analysis based on the attributes of the requests. It first filters out requests against the rules by rule matching. Then, the introspective variational autoencoder (IntroVAE) is used for behavior analysis to realize dynamic access decisions. Requests classified as normal can be authorized for access. ABAC-IntroVAE only needs samples of normal requests for training, avoiding the difficult task of collecting massive and frequently changing samples of attack requests. Meanwhile, the IntroVAE model is updated through continual learning to adapt to new-style normal behaviors due to the update of network services. Our experiment study suggests that our proposed ABAC-IntroVAE can effectively perform dynamic access control. It achieves an accuracy of 97.2% in abnormal detection and maintains an accuracy of over 97% through continual learning, despite the addition of new-style user behavior patterns.

Deep learning has made significant break through's in image classification, but accurate diagnosis on chest radiography remains challenging due to a variety of potential diseases contained in one scan. Complex relations among diseases have significant clinical meanings, but are always ignored in most of previous work. Thus in this paper, we propose a novel Label Generation Network (LGN) which treats the label sequence as the caption of a radiology image and utilizes RNN to generate the disease labels according to the semantic relations and co-occurrence dependency among them. However, the sequential generation process of RNN makes it hard to capture the complex topological relations among diseases. To mitigate this problem, a Historical Information Module (HIM) is especially introduced to LGN, in which all the generated labels are fully considered when generating a new label. Moreover, a specific self-attention mechanism is applied in HIM to learn the topological disease relations and utilize them to select useful historical information which can provide positive guidance to the prediction of new label. Very positive results have been obtained in our experiments on the benchmark dataset of Chest X-ray14, which significantly outperform the state-of-the-art methods.