Adversarial reprogramming allows repurposing a machine-learning model to perform different task. For example, trained recognize animals can be reprogrammed digits by embedding an adversarial program in the digit images provided as input. Recent work has shown that may not only used abuse models service, but also beneficially, improve transfer learning when training data is scarce. However, factors affecting its success are still largely unexplained. In this work, we develop first-order...

Insider threats have shown their great destructive power in information security and financial stability received widespread attention from governments organizations. Traditional intrusion detection systems fail to be effective insider attacks due the lack of extensive knowledge for behavior patterns. Instead, a more sophisticated method is required deeper understanding activities that insiders communicate with system. In this paper, we design classifier, neural network model utilizing Long...

High false-positive (FP) rate remains to be one of the major problems solved in CAD study because too many false-positively cued signals will potentially degrade performance detecting true-positive regions and increase call-back environment. In this paper, we proposed a novel classification method for FP reduction, where conventional "hard" decision classifier is cascaded with "soft" objective reduce false-positives cases multiple FPs retained after classification. The takes competitive...

The heading stage of maize is an important period during its growth and development indicates the beginning pollination. In this regard, automated method for tassel detection highly to monitor growth. However, recognition mainly relies on visual evaluation. This presents some limitations, such as expensive subjective. work proposed a novel automatic detection. algorithm, color attenuation prior model was used scene depth saturation graph remove image saturation. An Itti attention algorithm...

Traditional Chinese Medicine (TCM) is an influential form of medical treatment in China and surrounding areas. In this paper, we propose a TCM prescription generation task that aims to automatically generate herbal medicine based on textual symptom descriptions. Sequence-to-sequence (seq2seq) model has been successful dealing with sequence tasks. We explore potential end-to-end solution the using seq2seq models. However, experiments show directly applying leads unfruitful results due...

Traditional Chinese Medicine (TCM) has accumulated a big amount of precious resource in the long history development. TCM prescriptions that consist herbs are an important form treatment, which similar to natural language documents, but weakly ordered fashion. Directly adapting modeling style methods learn embeddings can be problematic as not strictly order, front prescription connected very last ones. In this paper, we propose represent with distributed representations via Prescription...

An algorithm is developed for fast, accurate identification of lung fields in chest radiographs use various computer- aided diagnosis (CAD) schemes. The method we presented simplifies the current approach edge detection from derivatives by using only first derivative profiles each image, and combining it with pattern classification image feature analysis determining both region interest (ROI) actual boundaries. Moreover, instead traditional curve fitting to delineate detected field, applied...

We present a novel Balanced Incremental Model Agnostic Meta Learning system (BI-MAML) for learning multiple tasks. Our method implements meta-update rule to incrementally adapt its model new tasks without forgetting old Such capability is not possible in current state-of-the-art MAML approaches. These methods effectively tasks, however, suffer from 'catastrophic forgetting' phenomena, which that are streamed into the degrade performance of on previously learned performs meta-updates with...

Most high-performance semantic segmentation networks are based on complicated deep convolutional neural networks, leading to severe latency in real-time detection. However, the state-of-the-art with low complexity still far from detecting objects accurately. In this paper, we propose a network, RecepNet, which balances accuracy and inference speed well. Our network adopts bilateral architecture (including detail path, path aggregation module). We devise lightweight baseline for gather rich...

法律惩罚的正当性是惩罚理论研究的永恒主题，如何系统性地反思并阐述惩罚的意义构成了不同立场下学者们的研究目标。然而相较于法律实证主义的研究，近现代自然法对于该问题的贡献却功薄蝉翼。这种情况一直到墨菲提出以自然法的立场展开对该问题的思考。他以自然法中的共同善作为惩罚理论的起点，不仅强调了惩罚理论在自然法中的研究价值，还总结并犀利点评了不同阵营的观点，并且在这些观点的基础上创新性地构建了一套基于自然法立场的报应主义惩罚理论。他着重强调的是法律惩罚作为一种次要回应共同善手段的合理性，即使对惩罚的预防功能讨论有所欠缺，也依然展现了墨菲法律惩罚理论的意义与价值。

Adversarial reprogramming allows repurposing a machine-learning model to perform different task. For example, trained recognize animals can be reprogrammed digits by embedding an adversarial program in the digit images provided as input. Recent work has shown that may not only used abuse models service, but also beneficially, improve transfer learning when training data is scarce. However, factors affecting its success are still largely unexplained. In this work, we develop first-order...

In this paper, we propose a novel method with Kolmogorov entropy to extract the feature of event-related EEG. The results show that can effectively quantify dynamic process desynchronization/synchronization (ERD/ERS) time course EEG in relation hand movement imagination. relative increase and decrease could be an indicator ERD/ERS. To testify validity measure, is tested on five human subjects for extraction classify left- right-hand motor imagery by Support Vector Machine (SVM) classifier....

Video prediction yields future frames by employing the historical and has exhibited its great potential in many applications, e.g., meteorological prediction, autonomous driving. Previous works often decode ultimate high-level semantic features to without texture details, which deteriorates quality. Motivated this, we develop a Pair-wise Layer Attention (PLA) module enhance layer-wise dependency of feature maps derived from U-shape structure Translator, coupling low-level visual cues...

Breast cancer is the second leading cause of death among American women. Computer aided diagnosis (CAD) has been proposed as a "second-opinion" strategy for breast screening. This paper presents new mammographic image processing method using anisotropic diffusion filtering to improve feature enhancement and lesion detection CAD in digital mammography.

Time series change point detection can identify the locations of abrupt points in many dynamic processes. It help us to find anomaly data an early stage. At same time, detecting for long, periodic, and multiple input has received a lot attention recently, is universally applicable fields including power, environment, finance, medicine. However, performance classical methods typically scales poorly such time series. In this paper, we propose CPMAN, novel prediction-based approach via...