Search

A continual learning (CL) model is designed to solve the catastrophic forgetting problem, which damages the performance of neural networks by overwriting previous knowledge with new knowledge. The fundamental cause of this problem is that previous data is not available when training new data in the CL setting. The memory-based CL methods leverage a memory buffer to address this problem by storing a limited subset of previous data for replay, and most methods of this type adopt random storage and replay strategies. In the human brain, the hippocampus replays consolidated knowledge from the neocortex in a random manner, e.g., random dreaming. Inspired by this memory mechanism, we propose a memory-based method, which replays more consolidated memory data while maintaining the randomness. Our work highlights that random replaying is important for the CL model, which confirms the effectiveness of random dreaming in the human brain.

This paper introduces the technical transformation and system realization of the hoister gap monitoring and protection device in Hongyang No.2 Coal Mine. Through the high-precision laser displacement sensor, we detected displacement signals, such as brake clearance, brake wear value, spring fatigue and brake disc deflection, to convert into 4-20 mA current signal to transmit to the PLC analog input port. PLC can read, analyze and convert the data in real time, and convert it into numerical value which is easy to display. Sound and light alarm signal is sent when the set value is exceeded. The practical application proves that the device has the technical characteristics of high sensitivity, strong anti-interference ability and reliable operation.

Continual learning (CL) from infinite data streams has become a challenge for neural network models in real-world scenarios. Catastrophic forgetting of previous knowledge occurs in this learning setting, and existing supervised CL methods rely excessively on accurately labeled samples. However, the real-world data labels are usually misled by noise, which influences the CL agents and aggravates forgetting. To address this problem, we propose a method named noise-labeled online continual learning (NLOCL), which implements the online CL model with noise-labeled data streams. NLOCL uses an empirical replay strategy to retain crucial examples, separates data streams by small-loss criteria, and includes semi-supervised fine-tuning for labeled and unlabeled samples. Besides, NLOCL combines small loss with class diversity measures and eliminates online memory partitioning. Furthermore, we optimized the experience replay stage to enhance the model performance by retaining significant clean-labeled examples and carefully selecting suitable samples. In the experiment, we designed noise-labeled data streams by injecting noisy labels into multiple datasets and partitioning tasks to simulate infinite data streams realistically. The experimental results demonstrate the superior performance and robust learning capabilities of our proposed method. [ABSTRACT FROM AUTHOR]

Adaptive laboratory evolution (ALE) is a technique for the selection of strains with better phenotypes by long-term culture under a specific selection pressure or growth environment. Because ALE does not require detailed knowledge of a variety of complex and interactive metabolic networks, and only needs to simulate natural environmental conditions in the laboratory to design a selection pressure, it has the advantages of broad adaptability, strong practicability, and more convenient transformation of strains. In addition, ALE provides a powerful method for studying the evolutionary forces that change the phenotype, performance, and stability of strains, resulting in more productive industrial strains with beneficial mutations. In recent years, ALE has been widely used in the activation of specific microbial metabolic pathways and phenotypic optimization, the efficient utilization of specific substrates, the optimization of tolerance to toxic substance, and the biosynthesis of target products, which is more conducive to the production of industrial strains with excellent phenotypic characteristics. In this paper, typical examples of ALE applications in the development of industrial strains and the research progress of this technology are reviewed, followed by a discussion of its development prospects. [ABSTRACT FROM AUTHOR]

By employing a site-specific mutagenesis strategy for glycerol kinase (encoding by glpK), the active-site residue Met270 was replaced with He in the Bacillus subtilis genome. A mutant strain Bacillus subtilis M270I was constructed successfully by using the markerless gene replacement system. The results of growth analysis indicated that the mutant strain got a higher level of glycerol utilization. Compared with the control strain Bacillus subtilis 168 Δupp, the mutant strain exhibited higher specific growth rate (increased by 11%) and higher biomass yield (improved by 16%) with a shorten lag phase (shorten 2~4 h) in the M9 glycerol minimal medium. [ABSTRACT FROM AUTHOR]

The effect of different microorganisms (lactic acid bacteria RG, lactic acid bacteria MSA, lactic acid bacteria FL, Bacillus KC, Bacillus ND, Bacillus JD) on the pH value and cell wall components (pectin, cellulose, hemicellulose and lignin) of tobacco stem was studied by enhancing the solid-state anaerobic fermentation of tobacco stem. The results showed that solid-state anaerobic fermentation could reduce the pH value of tobacco stem to varying degrees, and the pH value of tobacco stem fermented by lactic acid bacteria was reduced from 6.05 to 4.50 ~ 4.60. After 4 days of fermentation, the degradation effect of lactic acid bacteria RG on pectin degradation was the most obvious and the degradation rate of pectin in tobacco stem reached 87.75%. The degradation effect of lactic acid bacteria MAS on cellulose was the most obvious and the cellulose quality score decreased from 20.83% to 15.90%. Bacillus JD had better lignin and hemicellulose decomposition effect and the quality score of lignin and hemicellulose decreased from 10.91% and 10.66% to 5.55% and 7.40%, respectively. [ABSTRACT FROM AUTHOR]

Due to the contradiction between the discontinuity of agricultural production and continuity of industrial processing, the collection, storage and transformation of lignocellulosic biomass has become a basic problem in ensuring its sustainable development and large-scale utilization. Two broad methodologies for storage of LCB, dry and wet storage, are introduced. Process parameters including chemical reactions, microbial growth, temperature, and oxygen content, are analyzed in silage of different stages and with different microbial silage additives for the solid anaerobic treatment. Based on the results, we propose a quality evaluation system, encompassing sensory evaluation, organic acid content, soluble carbohydrate content, pH value and bacterial diversity, which can be used as criteria for judging the success of the microbial silage process. Finally, the research on solid-state anaerobic microbial silage pretreatment in biogas and ethanol conversion is summarized, and an optimized method for highly-efficient biochemical transformation of LCB is proposed based on the available findings. [ABSTRACT FROM AUTHOR]

The improved nutritional composition of anaerobic solid‐state fermentation (SSF) with broken dried chestnut by probiotic lactic acid bacteria (LAB), namely Lactobacillus casei strain CU (CU) and Lactobacillus fermentum strain KF5 (KF5) was evaluated. Preliminary sensory evaluation results showed that fermented chestnut had unique flavor characteristics between perception of sour and of chestnut. The number of LABs in fermented chestnut samples was not less than 8 log10 cfu/g under the condition of 40 days storage at 4°C. Chestnut fermented with KF5 had a slightly better antioxidant activity to reduce the 1,1‐Diphenyl‐2‐picrylhydrazyl (DPPH) free radical of 87% than the unfermented control (70.12%). Significant increases of 10.53% and 21.16% in protein digestibility of samples were observed following SSF with CU and KF5, respectively. The extracts of the fermented chestnut showed a less intense band just below 25 kDa in the sodium dodecyl sulfate‐polyacrylamide gel electrophoresis (SDS‐PAGE). The gas chromatography‐mass spectrometer (GC‐MS) extracellular metabolite analysis showed that the complex macromolecules in chestnut was degraded into simple sugars, organic acids, amino acids, and fatty acids. The findings of the study provide the possibility of using fermented chestnuts as potential healthy functional food. Practical applications: At present, the application scope of modern chestnut preservation technology is not very wide, which is limited by cost and other factors. Therefore, the development of in‐depth processing technologies of chestnut is one of the important ways to have a positive influence on the chestnut market. Nowadays, people focus on functional foods that can give full play to the body's defensive function, regulate physiological rhythm, prevent diseases, and promote rehabilitation, rather than just satisfying appetite or nutrition. Consumer attention to consume probiotic foods contributing to the maintenance of a good state of health through their biologically active ingredients has been significantly increased. This topic combines the characteristic resources of chestnut with selected probiotic lactic acid bacteria (LAB) through food biotechnology and uses anaerobic solid‐state fermentation (SSF) technology to develop healthy functional food with high nutritional value and in line with people's consumption concept: chestnut products with good nutritional flavor and probiotic functional, which further improves the nutritional value of products. [ABSTRACT FROM AUTHOR]

Defatted rice bran was taken as material, by high temperature alpha amylase liquefaction, liquid fermentation, acid leaching, alkaline protease hydrolysis and spray drying process, the butanol, phytate, rice bran protein and rice bran dietary fiber were co-produced. The process conditions and physicochemical properties of the products were studied. The results showed that the extraction sequence suitable for defatted rice bran was DRB liquefaction at high temperature, butanal fermentation, phytate, extract rice bran protein and rice bran dietary fiber separation. The mass concentration of butanol in the fermentation broth was up to 14. 03 g/L. The extraction rate of phytate was 6. 20%, while the content of P2O5 was 39. 77%, which fully met the standard of Chinese Pharmacopoeia. The extraction rate of rice bran protein was 12. 11%, while the content of total amino acids was 63. 32%. The composition of essential amino acids was comparable to that of egg protein, which had low allergy and was suitable for use in children's food. The extraction rate of rice bran dietary fiber was 25. 00%, while the content of dietary fiber was 72. 00%, which could be added to various health foods as functional food additives. [ABSTRACT FROM AUTHOR]

The effects of acid treatment, alkali treatment and microwave treatment on the enzymatic hydrolysis of tobacco stalks were studied by using tobacco processing by-product tobacco rods as raw materials, and the changes of composition and microstructure of tobacco stalks before and after treatment were analyzed. The results showed that the alkali treatment had a significant effect on the conversion efficiency of tobacco stalks in the three pretreatment methods. The tobacco stalks were treated with a mass fraction of 1% NaOH solution at 75 °C for 120 min, and the glucose content in the hydrolysate was 334. 69 mg/g tobacco stalks increased by 118. 37% compared to the control sample. The composition analysis showed that the degradation of tobacco lignin by alkali treatment was more obvious, and the degradation rate was 46. 45%. Alkali treatment not only destroyed the tobacco lignin benzene ring skeleton, but also modified the benzene ring side chain group, resulting in the disintegration of the lignin macromolecular structure, and the tobacco stalks enzymatic resistance barrier function was reduced, thereby improving the conversion efficiency of tobacco stalks. Alkali treatment did not change the crystallinity of tobacco stalks cellulose, and the increase in crystallinity caused by microwave and acid treatment would adversely affected the efficiency of enzymatic hydrolysis of tobacco stalks. [ABSTRACT FROM AUTHOR]