Robust training of median dendritic artificial neural networks for time series forecasting

Expert Systems with Applications, Journal Year: 2023, Volume and Issue: 238, P. 122080 - 122080

Published: Oct. 10, 2023

Language: Английский

---

Health indicators for remaining useful life prediction of complex systems based on long short-term memory network and improved particle filter

Reliability Engineering & System Safety, Journal Year: 2023, Volume and Issue: 241, P. 109666 - 109666

Published: Sept. 15, 2023

Language: Английский

---

A new deep neural network for forecasting: Deep dendritic artificial neural network

Artificial Intelligence Review, Journal Year: 2024, Volume and Issue: 57(7)

Published: June 11, 2024

Abstract Deep artificial neural networks have become a good alternative to classical forecasting methods in solving problems. Popular deep classically use additive aggregation functions their cell structures. It is available the literature that of multiplicative shallow produces successful results for problem. A type high-order network uses dendritic neuron model network, which has performance. In this study, transformation turned into multi-output architecture. new based on and proposed. The training carried out with differential evolution algorithm. performance compared basic some recent over stock market time series. As result, it been observed very

Language: Английский

---

A new global sine cosine algorithm for solving economic emission dispatch problem

Information Sciences, Journal Year: 2023, Volume and Issue: 648, P. 119569 - 119569

Published: Aug. 19, 2023

Language: Английский

---

Improved dendritic learning: Activation function analysis

Information Sciences, Journal Year: 2024, Volume and Issue: 679, P. 121034 - 121034

Published: June 15, 2024

Language: Английский

---

Application of artificial intelligence in coal mine ultra-deep roadway engineering—a review

Artificial Intelligence Review, Journal Year: 2024, Volume and Issue: 57(10)

Published: Aug. 19, 2024

The deep integration of computer field and coal mining is the only way to mine intellectualization. A variety artificial intelligence tools have been applied in open-pit shallow mines. However, with geometric increase demand, contradiction between supply demand becoming more serious, exploitation resources from layer (> 600 m) has become an inevitable trend. Well then, as a new engineering scene, harsh conditions "three high one disturbance" seriously threaten safety personnel. superposition complex environment makes number input factors sharply, which leads application roadway engineering. guidance not mature, construction various databases missing, there are still some problems universality applicability. To this end, paper starts introduction operating characteristics tools, conducts comprehensive study relevant high-level articles published top journals. It systematically sorts out research progress that successfully solved five directions rock mechanics strength, surrounding stability, rock-burst, roof fall risks micro-seismic events. While objectively evaluating performance different it also expounds its own views on key results. Literature review shows whether development tool or comparative model, ANN than 98%, performs extremely well direction stability risk, accuracy rate 90%. As most mature AI application, mechanical strength experienced process "SVM → DL XGBoost RF". dataset small samples (< 100) big 1000), R2 tree-based models can be stabilized at 95%. rock-burst prediction mainly focuses monitoring data. Whether sample large-scale data BN remains above 85%. evaluation events recent years. image processing CNN important. signal recognition classification accounts for 90%, potential source location needs further explored. In general, nature itself first choice almost all influencing factors. At same time, update iteration methods (micro-seismic, ground sound, separation, deformation, etc.) expands database, making possible obtain due threat life cost equipment, very difficult before. parameter selection method combining lithology conditions, geological will gradually research. Finally, follow-up work collation on-the-spot investigation, existing mines, explores engineering, puts forward focus challenging future, gives opinions.

Language: Английский

---

Evaluation of QSAR models for predicting mutagenicity: outcome of the Second Ames/QSAR international challenge project

SAR and QSAR in environmental research, Journal Year: 2023, Volume and Issue: 34(12), P. 983 - 1001

Published: Dec. 2, 2023

Quantitative structure-activity relationship (QSAR) models are powerful in silico tools for predicting the mutagenicity of unstable compounds, impurities and metabolites that difficult to examine using Ames test. Ideally, Ames/QSAR regulatory use should demonstrate high sensitivity, low false-negative rate wide coverage chemical space. To promote superior model development, Division Genetics Mutagenesis, National Institute Health Sciences, Japan (DGM/NIHS), conducted Second International Challenge Project (2020-2022) as a successor First (2014-2017), with 21 teams from 11 countries participating. The DGM/NIHS provided curated training dataset approximately 12,000 chemicals trial 1,600 chemicals, each participating team predicted various models. then test results assist improvement. Although overall performance on was not First, eight both projects achieved higher sensitivity than only Project. Thus, these evaluations have facilitated development QSAR

Language: Английский

---

Hybrid Sine Cosine and Particle Swarm Optimization Algorithm for High-Dimensional Global Optimization Problem and Its Application

Mathematics, Journal Year: 2024, Volume and Issue: 12(7), P. 965 - 965

Published: March 24, 2024

Particle Swarm Optimization (PSO) is facing more challenges in solving high-dimensional global optimization problems. In order to overcome this difficulty, paper proposes a novel PSO variant of the hybrid Sine Cosine Algorithm (SCA) strategy, named Velocity Four (VFSCPSO). The introduction SCA strategy velocity formulation ensures that optimal solution found accurately. It increases flexibility PSO. A series experiments are conducted on CEC2005 test suite with compositional algorithms, algorithmic variants, and good intelligent algorithms. experimental results show algorithm effectively improves overall performance algorithms; Friedman proves has competitiveness. also performs better PID parameter tuning. Therefore, VFSCPSO able solve problems way.

Language: Английский

---

Speech Emotion Recognition using Hybrid Architectures

International Journal of Computing, Journal Year: 2024, Volume and Issue: unknown, P. 1 - 10

Published: April 1, 2024

The detection of human emotions from speech signals remains a challenging frontier in audio processing and human-computer interaction domains. This study introduces novel approach to Speech Emotion Recognition (SER) using Dendritic Layer combined with Capsule Network (DendCaps). A Convolutional Neural (NN) Long Short-Time (CLSTM) hybrid model are used create baseline which is then compared the DendCap model. Integrating dendritic layers capsule networks for emotion can harness unique advantages both architectures, potentially leading more sophisticated accurate models. layers, inspired by nonlinear properties trees biological neurons, handle intricate patterns variabilities inherent signals, while networks, their dynamic routing mechanisms, adept at preserving hierarchical spatial relationships within data, enabling capture refined emotional subtleties speech. main motivation DendCaps bridge gap between capabilities neural systems artificial networks. combination aims capitalize on nature where dependencies be better captured. Finally, two ensemble methods namely stacking boosting evaluating CLSTM experimental results show that gives superior result 75% accuracy.

Language: Английский

---

Grey Wolf Optimization Algorithm Based on Follow-Controlled Learning Strategy

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 101852 - 101872

Published: Jan. 1, 2023

This paper analyzes the OBL strategy's impact on optimizing GWO algorithm and identifies three shortcomings. specific limitations of optimization approach. To address these shortcomings enhance both global local exploration capabilities GWO, this introduces a follow-controlled opposition learning strategy. then, control parameter C grey wolf to investigate its exploration. Based properties, new is proposed. The proposed strategy are introduced into traditional obtain FCGWO algorithm. Finally, conducts comparative analysis in comparison other meta-heuristic algorithms, as well enhanced algorithm, utilizing 23 benchmark test functions 2 engineering problems. results indicate that effectively avoids OBL, while also outperforming algorithms significantly terms solution quality.

Language: Английский

---