Inception-v3 with reduce learning rate for optimization of lung cancer histopathology classification
Bibliographic record
Abstract
Cancer is an uncontrolled and destructive proliferation of body cells.Lung cancer is the highest cause of death in Indonesia, especially among men.The cancer patient can be examined for histopathological checkups.This examination is carried out by taking body tissue at a place where cancer cells are suspected.Histopathology is the gold standard for detecting pathology or abnormalities in body cells.The results of a histopathological examination can differentiate between normal body cells, cancer cells, and their types.There are three types of lung cancer histopathology: adenocarcinomas, squamous cell carcinomas, and benign lung tissues.Classification of lung cancer using histopathology images is an alternative to detecting the severity of cancer.This study used Deep Learning Convolutional Neural Network (CNN).Transfer learning utilizes ImageNet weights and biases from the Inception-v3 pre-trained network, so a new model is not trained from scratch.The hyperparameter uses a learning rate (LR) of 0.0001, epoch 50, batch-size 32, and RMSProp optimization.In addition, there is tuning with reduced lr when there is an increase in validation loss before reaching the maximum epoch.The dataset uses the LC25000.The data consists of 3,000 images, three classes with 1,000 classes per class.The best results show accuracy, precision, and recall are 99.17%,99.17%, and 99%, respectively.Performance increased by 3% compared to the baseline method without learning rate tuning.
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 | 0.002 |
| 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 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.004 | 0.002 |
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".