Canada (2013)" COMPRESSIVE GAUSSIAN MIXTURE ESTIMATION
Bibliographic record
Abstract
Algorithms using sketches can be found in the database literature [2, 3]. In this case, a sketch is a compressed representation of the data and can be updated whenever an element is added or removed from the database without reprocessing all the data. A popular application is the search for frequent items in a data stream, also called heavy hitters [4]. Histogram fitting using a sketch has been explored [5] in the case of n-dimensional discrete vectors. In this case, the sketch is an accumulated random projection of the vectors. This method does not scale to high dimensions, the construction of the histogram from the sketch having complexity which is exponential in n. Inverse problems on density mixtures have also been studied [6, 7]. Given data drawn from a mixture of candidate functions, both papers propose to cast the reconstruction as the optimization of a sparsity-inducing cost function on the vector of mixture coefficients. Both methods require the considered set of candidate densities to be finite and the elements of this set to be incoherent, i.e., different from each other. These assumptions do not hold for GMM: the centroids of the Gaussians can vary continuously and be arbitrarily close to one another. Compressed representations of data vectors based on random projections for linear classification have been studied in [8, 9]. These compressed representations aim at replacing a vector x by Mx where M is a dimensionality-reducing matrix, but do not compress the whole data set to a size that is independent of the number N of vectors in the set. In [10], a compressed representation of sparse probability distributions over multidimensional binary vectors is studied. The comhal-00799896,
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 | 0.002 | 0.003 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.005 | 0.002 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.003 | 0.001 |
| Insufficient payload (model declined to judge) | 0.197 | 0.064 |
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".