Anomaly Detection and Approximate Similarity Searches of Transients in Real-time Data Streams
Bibliographic record
Abstract
This is the official Zenodo version of the code LAISS (Lightcurve Anomaly Identification and Similarity Search), associated with the paper, "Anomaly Detection and Approximate Similarity Searches of Transients in Real-time Data Streams" by Aleo et al (in review). This repository contains all datasets and code needed to run a local instance of LAISS, though slight modifications will be needed (e.g., renaming hard-coded file paths). See Aleo et al. for details on the LAISS pipeline, now on arXiv and currently submitted to The Astrophysical Journal. The live version of the code can be found on Github. Moreover, the results of all objects processed by LAISS via the ANTARES broker is available on the main page and selecting “LAISS_RFC_AD_filter” under ‘Tags’. Those we consider anomalies are objects that have a Locus Property feature “LAISS_RFC_anomaly_score” > 0.5. Note that the version on ANTARES has no similarity search functionality.A demo can be found on Google Colab, written by current code maintainer Alex Gagliano.Below we list the files with a brief description: "LAISS_ANNOY_pseudo_Filter.ipynb" -- The notebook version of LAISS. Preferred method because it doesn't need to reload the large .ann files for each instance. "LAISS.py" -- The .py version of LAISS. Same functionality but a little slower because of the many arguments and longer runtimes due to needing to reload the .ann files for each run. Can be run, e.g., withLAISS(l_or_ztfid_ref="ZTF18abydmfv", lc_and_host_features=lc_and_host_features, n=8, use_lc_for_ann_only_bool=True, use_ysepz_phot_snana_file=False, show_lightcurves_grid=False, show_hosts_grid=False, run_AD_model=False, savetables=False, savefigs=False) "*.ann" & "*.npy"-- The ANNOY index files, used for similarity search functionality."*.csv.gz" -- Datafiles with objects (rows) and light curve + host features (columns), used for anomaly detection and similarity search functionality. NOTE: With either choice of running LAISS, you'll need to add the following hardcoded directories (or manually change the filepaths). See Github for directory structure:tables/custom/timeseries/notebooks/ysepz_snana_phot_files/notebooks/LAISS_run/loci_dbs/alerce_cut/ps1_psc/ps1_cutouts/dataframes/RFC/SMOTE_train_test_70-30_min14_kneighbors8/cls=binary_n_estimators=100_max_depth=35_rs=11_max_feats=35_cw=balanced/figuresRFC/SMOTE_train_test_70-30_min14_kneighbors8/cls=binary_n_estimators=100_max_depth=35_rs=11_max_feats=35_cw=balanced/model
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 | 0.002 | 0.010 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.005 | 0.004 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.003 | 0.003 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.013 | 0.015 |
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".