A Comparison of Adaptive Moment Estimation (Adam) and RMSProp Optimisation Techniques for Wildlife Animal Classification Using Convolutional Neural Networks
Bibliographic record
Abstract
The rapid decline in wildlife animal diversity necessitates expedited evaluations of biodiversity and population dynamics. Accurate image recognition from camera traps is central to such assessments. This study investigates the impact of different optimisation techniques and hyperparameter configurations on the accuracy of wildlife animal classification. Specifically, the comparative effectiveness of the Adaptive Moment Estimation (Adam) and Root Mean Square Propagation (RMSProp) optimisation algorithms is examined. The influence of learning rates on these optimisation techniques is evaluated, while other hyperparameters are held constant. Convolutional Neural Networks (CNN) models, namely DenseNet-121, ResNet-50, and AlexNet, are utilised for this study. The investigation employs a dataset composed of 47,841 images sourced from the Serengeti Project Season 1 Snapshot in Tanzania. The images depict wild animals in diverse perspectives within their natural habitats, with some providing a complete view of the animal's body, while others do not. The dataset, characterised by an imbalanced distribution, is segregated into training, validation, and testing sets at proportions of 80%, 10%, and 10%, respectively. The results reveal that the application of the Adam optimisation technique yields the highest average accuracy of 80.66% with the ResNet-50 model. However, the DenseNet-121 model achieved an overall accuracy exceeding 95%. Notably, the ResNet-50 architecture, with learning rates of 0.1 and 0.01, encountered challenges during the training and validation of all images due to the complexity of the dataset. Irrespective of the optimisation technique employed, the most effective performance was observed with the ResNet-50 model, utilising the Adam optimiser and a learning rate of 0.001. The study proposes suitable learning rate values for training scenarios similar to the present investigation.
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 | 0.003 | 0.007 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 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".