MF-MNER: Multi-models Fusion for MNER in Chinese Clinical Electronic Medical Records.

To address the problem of poor entity recognition performance caused by the lack of Chinese annotation in clinical electronic medical records, this paper proposes a multi-medical entity recognition method F-MNER using a fusion technique combining BART, Bi-LSTM, and CRF. First, after cleaning, encoding, and segmenting the electronic medical records, the obtained semantic representations are dynamically fused using a bidirectional autoregressive transformer (BART) model. Then, sequential information is captured using a bidirectional long short-term memory (Bi-LSTM) network. Finally, the conditional random field (CRF) is used to decode and output multi-task entity recognition. Experiments are performed on the CCKS2019 dataset, with micro avg Precision, macro avg Recall, weighted avg Precision reaching 0.880, 0.887, and 0.883, and micro avg F1-score, macro avg F1-score, weighted avg F1-score reaching 0.875, 0.876, and 0.876 respectively. Compared with existing models, our method outperforms the existing literature in three evaluation metrics (micro average, macro average, weighted average) under the same dataset conditions. In the case of weighted average, the Precision, Recall, and F1-score are 19.64%, 15.67%, and 17.58% higher than the existing BERT-BiLSTM-CRF model respectively. Experiments are performed on the actual clinical dataset with our MF-MNER, the Precision, Recall, and F1-score are 0.638, 0.825, and 0.719 under the micro-avg evaluation mechanism. The Precision, Recall, and F1-score are 0.685, 0.800, and 0.733 under the macro-avg evaluation mechanism. The Precision, Recall, and F1-score are 0.647, 0.825, and 0.722 under the weighted avg evaluation mechanism. The above results show that our method MF-MNER can integrate the advantages of BART, Bi-LSTM, and CRF layers, significantly improving the performance of downstream named entity recognition tasks with a small amount of annotation, and achieving excellent performance in terms of recall score, which has certain practical significance. Source code and datasets to reproduce the results in this paper are available at https://github.com/xfwang1969/MF-MNER .

Genetic algorithm assisted bridge fiber design and fabrication for few-mode multi-core fiber Fan-in/Fan-out device.

We present a rapid and precise method to design the multiple step-index bridge fiber for ultra-low insertion loss few-mode multi-core fiber Fan-in/Fan-out device. The genetic algorithm is applied to optimize the structural parameters to support multi-mode operation. Based on the proposed intelligent iteration platform, core-based multiplex/demultiplex optimization can be achieved with less than 1.0 dB insertion loss for the first 6 LP modes in space division multiplexing system consisting of few-mode multi-core fibers. Besides, we have successfully drew the designed bridge fiber and fabricated the corresponding Fan-in/Fan-out device. When connecting it with the single-core 6-mode fiber and 7-core 6-mode fiber, the average insertion losses of mode LP01, LP11a, LP11b, LP21a, LP21b, and LP02 are 0.88 dB, 1.11 dB, 1.07 dB, 1.42 dB, 1.33 dB, and 1.04 dB, respectively.

24 km High-Performance Raman Distributed Temperature Sensing Using Low Water Peak Fiber and Optimized Denoising Neural Network.

Raman distributed optical fiber temperature sensing (RDTS) has been extensively studied for decades because it enables accurate temperature measurements over long distances. The signal-to-noise ratio (SNR) is the main factor limiting the sensing distance and temperature accuracy of RDTS. We manufacture a low water peak optical fiber (LWPF) with low transmission loss to improve the SNR for long-distance application. Additionally, an optimized denoising neural network algorithm is developed to reduce noise and improve temperature accuracy. Finally, a maximum temperature uncertainty of 1.77 °C is achieved over a 24 km LWPF with a 1 m spatial resolution and a 1 s averaging time.

Single-ended self-calibration high-accuracy Raman distributed temperature sensing based on multi-core fiber.

Raman distributed temperature sensing (RDTS) obtains the temperature information by measuring the intensities of Raman scattering lights. The anti-stokes only RDTS can avoid the error caused by wavelength-dependent loss and dispersion. However, to eliminate temperature-independent intensity variations, single-wavelength demodulation generally adopts the double-ended detection scheme. This requires two optical fibers or one fiber to be folded into a loop, which is inconvenient in practical applications. Moreover, the temperature accuracy of such a scheme is lower than the conventional single-ended system, so it has not been widely used. Here, we propose and experimentally demonstrate a multi-core fiber (MCF) based RDTS system. A single-ended loop structure is achieved by connecting two cores at the far end of the MCF with a fan-in/fan-out device. By measuring the backscattered anti-stokes lights in the two cores, the results can be self-calibrated to eliminate the influence of temperature-independent light intensity changes. Besides, the results can be improved by averaging the temperatures of the two cores due to the spatial consistency of the MCF. Moreover, to further improve the temperature uncertainty, we employ the one-dimensional denoising convolutional neural network. Finally, a maximum temperature uncertainty of 1.4 °C is achieved over a 10 km MCF with a 3 m spatial resolution.

Digital in-service relative time delay estimation for SDM self-homodyne coherent systems.

For space division multiplexing self-homodyne coherent systems, we propose a novel digital in-service relative time delay (RTD) estimation method without any additional optoelectronic device. Taking advantage of the frequency-domain periodicity of the colored frequency modulation noise, we manage to find the peak with location reflecting the RTD in its autocorrelation function (ACF). The peak to average ratio is further enhanced by leveraging a low-pass differential finite impulse response filter for robust identification. By simulations, the method is validated to be feasible for various linewidths, formats (16QAM, 32QAM and 64QAM), and links up to 80 km. Particularly, it is proved to be inherently compatible with large-linewidth low-cost lasers for the 10-km link. Also, for a low-complexity implementation, we discuss the way to reduce the number of points used to calculate the ACF while maintaining the same dynamic range. Furthermore, we demonstrate a 50-GBaud 16-QAM experiment to investigate its performances. With received optical power varying from -11 dBm to -17 dBm, 216 points are sufficient to provide an estimation accuracy of standard deviation (STD) less than 0.089 ns for the RTD range of [2.6, 491.0 ns]. The STD can be lowered to 0.036 ns by adopting 218 points. Especially, at -11-dBm ROP, the highest performance has been achieved with an accuracy smaller than the symbol period (0.018-ns STD) and a RTD range of [1.5, 491.0 ns].

Remarks on Computational Method for Identifying Acid and Alkaline Enzymes.

The catalytic efficiency of the enzyme is thousands of times higher than that of ordinary catalysts. Thus, they are widely used in industrial and medical fields. However, enzymes with protein structure can be destroyed and inactivated in high temperature, over acid or over alkali environment. It is well known that most of enzymes work well in an environment with pH of 6-8, while some special enzymes remain active only in an alkaline environment with pH > 8 or an acidic environment with pH < 6. Therefore, the identification of acidic and alkaline enzymes has become a key task for industrial production. Because of the wide varieties of enzymes, it is hard work to determine the acidity and alkalinity of the enzyme by experimental methods, and even this task cannot be achieved. Converting protein sequences into digital features and building computational models can efficiently and accurately identify the acidity and alkalinity of enzymes. This review summarized the progress of the digital features to express proteins and computational methods to identify acidic and alkaline enzymes. We hope that this paper will provide more convenience, ideas, and guides for computationally classifying acid and alkaline enzymes.

Online Fault Diagnosis for Biochemical Process Based on FCM and SVM.

Fault diagnosis is becoming an important issue in biochemical process, and a novel online fault detection and diagnosis approach is designed by combining fuzzy c-means (FCM) and support vector machine (SVM). The samples are preprocessed via FCM algorithm to enhance the ability of classification firstly. Then, those samples are input to the SVM classifier to realize the biochemical process fault diagnosis. In this study, a glutamic acid fermentation process is chosen as an example to diagnose the fault by this method, the result shows that the diagnosis time is largely shortened, and the accuracy is extremely improved by comparing to a single SVM method.