Toward Predictive Chemical Deformulation Enabled by Deep Generative Neural Networks
Bibliographic record
Abstract
The design of chemical formulations is a challenging, high-dimensional problem. In typical formulations, tens of thousands of ingredients are available for use, yet only a tiny fraction end up in a given formulation. Deformulation, the problem of reverse engineering the precise amounts of each ingredient starting from just a list of ingredients, is similarly challenging but is a key capability for staying up-to-date with industry competitors. Here, we take advantage of a large, curated formulations dataset from CAS, a division of the American Chemical Society, which offers a consistent and highly structured representation of the formulations and the chemical identities of their components to show that a variational autoencoder neural network learns meaningful representations of formulations in various product classes such as antiperspirants and oral care. Furthermore, it can be used in conjunction with a two-step sampling algorithm to generate accurate ingredient amount suggestions for deformulation. Deformulation using a variational autoencoder produces estimates that are significantly more accurate than nearest neighbor methods, extrapolates better to formulations that are significantly different than previously seen formulations, and provides a way to leverage large datasets for industrially relevant capabilities.
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 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| 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".