Backpropagation is a cornerstone algorithm in training neural networks for supervised learning, which uses a gradient descent method to update network weights by minimizing the discrepancy between actual and desired outputs. Despite its pivotal role in propelling deep learning advancements, the biological plausibility of backpropagation is questioned due to its requirements for weight symmetry, global error computation, and dual-phase training. To address this long-standing challenge, many studies have endeavored to devise biologically plausible training algorithms. However, a fully biologically plausible algorithm for training multilayer neural networks remains elusive, and interpretations of biological plausibility vary among researchers. In this study, we establish criteria for biological plausibility that a desirable learning algorithm should meet. Using these criteria, we evaluate a range of existing algorithms considered to be biologically plausible, including Hebbian learning, spike-timing-dependent plasticity, feedback alignment, target propagation, predictive coding, forward-forward algorithm, perturbation learning, local losses, and energy-based learning. Additionally, we empirically evaluate these algorithms across diverse network architectures and datasets. We compare the feature representations learned by these algorithms with brain activity recorded by non-invasive devices under identical stimuli, aiming to identify which algorithm can most accurately replicate brain activity patterns. We are hopeful that this study could inspire the development of new biologically plausible algorithms for training multilayer networks, thereby fostering progress in both the fields of neuroscience and machine learning.
Download This Paper Open PDF in Browser Add Paper to My Library Share: Permalink Using these links will ensure access to this page indefinitely Copy URL Copy DOI
Electricity is a basic necessary ingredient for modern economic growth and electricity sectors and also is the main sources of Carbon dioxides (CO 2 ) emission. As the second largest electricity producer and consumer in the world, China has electricity sector as one of the most important sector controlling CO 2 emission. This study presents the results of a life cycle analysis of CO 2 emissions from power generation systems in China using input-output analysis, from both electricity producer perspective and electricity consumer perspective. In 2007, direct CO 2 emission per KWh of electricity generated by power plants and consumed by end users in China was at 701.7 gCO 2 and 748.7 gCO 2 respectively, while the Life cycle CO 2 emission (total CO 2 emission) is at 890.4 gCO 2 per kWh of electricity generated and 977.2 gCO 2 per KWh of electricity end-users consume. Of the total, embodied CO 2 emission caused by indirect use of energy and materials contributes to 21.2 percent. China has achieved significant reduction of CO 2 emission per KWh electricity since 1980 and still has great potential in future through suitable CO 2 management measures adopted.
A series of orthogonal array tests and single-factor tests were designed to study the growth of Bacillus sp. H strain isolated from Microcystis aeruginosa in the Taihu Lake. The results indicated that within the experimental scope the factors affecting the growth of Bacillus sp. H strain were arranged in the order: organic carbontempraturephosphateammonium. Of the four factors, organic carbon and temperature most significantly affected the growth of Bacillus sp. H strain. It was also found that with the raising of temperature, concentration of organic carbon or the ratio of N/P, the growth of Bacillus sp. H strain would become better. The optimum conditions for the growth of Bacillus sp. H strain in this experiment were:temperature 30℃, concentration of ammonium 0.520 g/L , concentration of peptone 2.0 g/L , phosphate 0.088 g/L , and organic carbon 2.28 g/L. Fig 4, Tab 2, Ref 5
This paper focuses on improving fire suppression performance through the use of compressed-air launching technology. A launch dynamics calculation model of a compressed-air launcher is presented, developed using VC++ programming, to simulate the acceleration process of a fire-extinguishing bomb in a barrel. By analyzing the influences of various structural and initial parameters on interior ballistics variations, the effectiveness of the calculation model and program in accurately simulating the launching process is demonstrated. The calculation results indicate that the bore pressure follows a similar trend to that of traditional gunpowder launching. Additionally, it is found that specific structural parameters, such as nozzle diameter and gas cylinder volume, have a direct impact on interior ballistics variations. Notably, the nozzle diameter positively affects the peak pressure, muzzle velocity, gas transfer efficiency, and launch efficiency. To ensure an optimal launch effect and efficiency, the nozzle diameter should be selected to be more than half of the launcher caliber. Similarly, the gas cylinder volume positively influences the peak pressure and muzzle velocity while negatively affecting the gas transfer efficiency and launch efficiency. Furthermore, the initial pressure in the gas cylinder exhibits a positive linear relationship with both the peak pressure and muzzle velocity but a negative linear relationship with the gas transfer efficiency and launch efficiency. The loading position minimally impacts the peak pressure and muzzle velocity but slightly enhances the gas transfer efficiency and launch efficiency. Finally, it is observed that launch angles do not affect the interior ballistic process. The research findings provide valuable theoretical guidance for determining the working parameters of compressed-air accelerated fire-extinguishing bombs.
Elodea nuttalli,Potamogeton maackianus and Alternanthera philoxeroides were crushed,pressed and then filtrated to get filtrate of hydrophyte as fermentation material.Effects of varying levels of anaerobic sludge and additional N on liquid state fermentation were studied.The results show that highest biogas production(863.2 mL)in which CH_4 content was 720～750 mL·L~(-1) in produced biogas and COD_(Cr) removal rate(90%) were reached within 10 days after 200 mL·L~(-1) of anaerobic sludge was added.Biogas production from anaerobic fermentation of press filtrate of Elodea nuttalli,Potamogeton maackianus and Alternanthera philoxeroides was increased by 2%,10% and 30%,respectively,when 1 g·kg~(-1) N was (added).
Abstract In order to monitor and evaluate the impact of rocket launching shock, strain, vibration, typhoon, earthquake and other loads on the launch tower, an online monitoring and evaluation system was established based on multi-sensor information fusion technology in a space launch site. Firstly, the system integrates the information of vibration, pressure, deformation, wind load, temperature and other sensors to comprehensively analyze and evaluate the safety of tower structure, and realizes health diagnosis. Then use 3dmax to draw the 3D simulation model, and realize the real-time 3D display of various monitoring and analysis results based on the 3D visualization and data-driven technologies.
Abstract Mammalian genomes are folded by the distinct actions of SMC complexes which include the chromatin loop-extruding cohesin, the sister-chromatid cohesive cohesin, and the mitotic chromosome-associated condensins. While these complexes function at different stages of the cell cycle, they co-exist on chromatin during the G2/M-phase transition, when genome structure undergoes a dramatic reorganization. Yet, how distinct SMC complexes affect each other and how their mutual interplay orchestrates the dynamic folding of 3D genome remains elusive. Here, we engineered all possible cohesin/condensin configurations on mitotic chromosomes to delineate the concerted, mutual influential action of SMC complexes. We find that: (i) The mitotic SMC complex condensin disrupts the focal accumulation of extrusive-cohesin at CTCF binding sites, thereby promoting the disassembly of interphase TADs and chromatin loops during mitotic progression. Conversely, extrusive-cohesin can impair condensin activity and alter mitotic chromosome helicity. (ii) Condensin diminishes cohesive-cohesin focal enrichment and, conversely, cohesive-cohesin can counteract condensin function and impede mitotic chromosome longitudinal shortening. (iii) The co-presence of extrusive- and cohesive-cohesin synergistically antagonizes condensin function and dramatically delays mitotic chromosome condensation. (iv) Extrusive-cohesin positions cohesive-cohesin at CTCF binding sites. However, cohesive-cohesin by itself is insufficient to mediate the formation of TADs or chromatin loop, implying non-overlapping function with extrusive-cohesin. Instead, cohesive-cohesin restricts chromatin loop expansion, potentially by limiting extrusive-cohesin movement. Collectively, our data describe a comprehensive three-way interplay among major SMC complexes that dynamically sculpts chromatin architecture during cell cycle progression.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)