Experimental Analysis of Training Parameters Combination of ANN Backpropagation for Climate Classification
Bibliographic record
Abstract
Artificial Neural Networks are widely used in prediction activities and classification processes. However, the implementation on average only uses a network architecture with one hidden layer, while the development of architectures with two or three hidden layers has not been done much. This article discusses the process of developing ANN Backpropagation using a Matlab-based graphical user interface with three hidden layers combined with non-linear activation functions (logsig, tansig, tanh) and training functions (trainrp and trainlm) based on learning rate and momentum. Architecture was created to study climate change in the area around Lombok International Airport station by training on hydrological data (rainfall) from January 2012 to December 2021 with a data type of 10-day interval (36 data every year). The number of neurons in the first hidden layer was determined using the Hecht-Nielsen model, while the second and third hidden layers used the Lawrence-Fredrickson model. Simulation results with architecture 36-73-37-19-1, a learning rate of 0.1, and momentum of 0.9 showed that variations in the activation function logsig-logsig-logsig-purelin and trainlm function demonstrated the best result with epoch of 7, MSE of 0.00090, and RMSE of 0.03011 in the training process and epoch of 5, MSE of 0.003758, and RMSE of 0.0613 in the data testing process. Furthermore, the prediction results demonstrated that a Q-value of 0.222 based on the Schmidt-Ferguson criteria obtained higher rainfall intensity information than previous years with climate category B (wet). Therefore, the government must be careful in determining policies related to field activities especially in agriculture because of climatic conditions with high rainfall.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".