A Comparative Study on the Training Effects of Different Optimizers for Deep Learning Models
Bibliographic record
Abstract
The training efficiency and generalization performance of deep learning models are highly dependent on the selection of optimizers. Differences in gradient update strategies among various optimizers directly affect the model's convergence speed, final accuracy, and training stability. Taking the house price prediction task as the research carrier, this paper constructs a fully connected neural network model based on the Boston Housing Dataset to systematically compare the training effects of three classic optimizers: Stochastic Gradient Descent (SGD), Adaptive Moment Estimation (Adam), and Root Mean Square Propagation (RMSprop). By controlling irrelevant variables such as model structure, learning rate, and batch size, quantitative analysis is conducted from three core dimensions: convergence speed, final prediction accuracy, and training stability. The applicable scenarios of each optimizer are discussed in combination with experimental results. Experiments show that the Adam optimizer has the fastest convergence speed and can quickly reduce the loss value in the early stage of training; the SGD optimizer, although converging slowly, can achieve the optimal final prediction accuracy after sufficient training; the RMSprop optimizer achieves a balance between convergence speed and stability, making it suitable for scenarios with non-stationary objective functions. The research results can provide practical references for optimizer selection in deep learning regression tasks, helping to improve the efficiency and performance of model training.
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 | 0.005 | 0.013 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
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".