Exemplar VAE: Linking Generative Models, Nearest Neighbor Retrieval, and\n Data Augmentation
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
We introduce Exemplar VAEs, a family of generative models that bridge the gap\nbetween parametric and non-parametric, exemplar based generative models.\nExemplar VAE is a variant of VAE with a non-parametric prior in the latent\nspace based on a Parzen window estimator. To sample from it, one first draws a\nrandom exemplar from a training set, then stochastically transforms that\nexemplar into a latent code and a new observation. We propose retrieval\naugmented training (RAT) as a way to speed up Exemplar VAE training by using\napproximate nearest neighbor search in the latent space to define a lower bound\non log marginal likelihood. To enhance generalization, model parameters are\nlearned using exemplar leave-one-out and subsampling. Experiments demonstrate\nthe effectiveness of Exemplar VAEs on density estimation and representation\nlearning. Importantly, generative data augmentation using Exemplar VAEs on\npermutation invariant MNIST and Fashion MNIST reduces classification error from\n1.17% to 0.69% and from 8.56% to 8.16%.\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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.002 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.004 |
| Open science | 0.004 | 0.011 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 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 it