Further Improvements to AngryTops: A machine learning package for the reconstruction of top-quark pair decay kinematics
Bibliographic record
Abstract
Studies of the top quark provide unique insights into the Standard Model due to its large mass. However, the kinematics of $t\bar{t}$ decays is difficult to reconstruct due to the complexity of these events and limited detector resolution. Neural networks are thought to perform as well as state-of-the-art statistical algorithms for reconstruction purposes. Our group has developed a machine learning package called AngryTops, a BLSTM neural network that reconstructs $t\Bar{t}$ decay pair kinematics resulting from 13 TeV $pp$ collisions. Although the package successfully reconstructs the kinematic variable distributions, we lack a systematic way to evaluate the network's performance on individual events. We implement improvements to better characterize the network's performance. We also introduce an algorithm that matches the observed leptons and jets with particles arising from the $t\bar{t}$ decay (truth particles). The variables used for matching are then used to filter the training dataset, retaining only events that the network should be able to reconstruct well. We train the network on the filtered dataset and evaluate how the network performs compared to when it is trained on the original sample. We also develop a $\chi^2$ metric and corresponding p-value test to assess predictions for each event. Further developments including data augmentation and fine-tuning parameters will be investigated using the $\chi^2$ test and matching algorithm as performance metrics.
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 | 0.003 | 0.008 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.003 |
| Open science | 0.004 | 0.002 |
| Research integrity | 0.001 | 0.003 |
| Insufficient payload (model declined to judge) | 0.031 | 0.013 |
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".