Progressive Gradient Pruning for Classification, Detection and\n DomainAdaptation
Bibliographic record
Abstract
Although deep neural networks (NNs) have achievedstate-of-the-art accuracy in\nmany visual recognition tasks,the growing computational complexity and energy\ncon-sumption of networks remains an issue, especially for ap-plications on\nplatforms with limited resources and requir-ing real-time processing. Filter\npruning techniques haverecently shown promising results for the compression\nandacceleration of convolutional NNs (CNNs). However, thesetechniques involve\nnumerous steps and complex optimisa-tions because some only prune after\ntraining CNNs, whileothers prune from scratch during training by\nintegratingsparsity constraints or modifying the loss function.In this paper we\npropose a new Progressive GradientPruning (PGP) technique for iterative filter\npruning dur-ing training. In contrast to previous progressive\npruningtechniques, it relies on a novel filter selection criterion thatmeasures\nthe change in filter weights, uses a new hard andsoft pruning strategy and\neffectively adapts momentum ten-sors during the backward propagation pass.\nExperimentalresults obtained after training various CNNs on image datafor\nclassification, object detection and domain adaptationbenchmarks indicate that\nthe PGP technique can achievea better trade-off between classification accuracy\nand net-work (time and memory) complexity than PSFP and otherstate-of-the-art\nfilter pruning techniques.\n
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) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| 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".