| 1. |
- Acharya, U.A., et al.
(author)
-
Systematic study of nuclear effects in p+Al, p+Au, d+Au, and He 3 + Au collisions at sNN =200 GeV using π0 production
- 2022
-
In: Physical Review C. - 2469-9985. ; 105:6
-
Journal article (peer-reviewed)abstract
- The PHENIX Collaboration presents a systematic study of inclusive π0 production from p+p, p+Al, p+Au, d+Au, and He3+Au collisions at sNN=200GeV. Measurements were performed with different centrality selections as well as the total inelastic, 0-100%, selection for all collision systems. For 0-100% collisions, the nuclear-modification factors, RxA, are consistent with unity for pT above 8GeV/c, but exhibit an enhancement in peripheral collisions and a suppression in central collisions. The enhancement and suppression characteristics are similar for all systems for the same centrality class. It is shown that for high-pT-π0 production, the nucleons in the d and He3 interact mostly independently with the Au nucleus and that the counterintuitive centrality dependence is likely due to a physical correlation between multiplicity and the presence of a hard scattering process. These observations disfavor models where parton energy loss has a significant contribution to nuclear modifications in small systems. Nuclear modifications at lower pT resemble the Cronin effect - an increase followed by a peak in central or inelastic collisions and a plateau in peripheral collisions. The peak height has a characteristic ordering by system size as p+Au>d+Au>He3+Au>p+Al. For collisions with Au ions, current calculations based on initial-state cold nuclear matter effects result in the opposite order, suggesting the presence of other contributions to nuclear modifications, in particular at lower pT. © 2022 American Physical Society.
|
|
| 3. |
- Matsoukas, Christos, et al.
(author)
-
Adding seemingly uninformative labels helps in low data regimes
- 2020
-
In: 37th International Conference on Machine Learning, ICML 2020. - : International Machine Learning Society (IMLS). ; , s. 6731-6740
-
Conference paper (peer-reviewed)abstract
- Evidence suggests that networks trained on large datasets generalize well not solely because of the numerous training examples, but also class diversity which encourages learning of enriched features. This raises the question of whether this remains true when data is scarce - is there an advantage to learning with additional labels in low-data regimes In this work, we consider a task that requires difficult-To-obtain expert annotations: Tumor segmentation in mammography images. We show that, in low-data settings, performance can be improved by complementing the expert annotations with seemingly uninformative labels from non-expert annotators, turning the task into a multi-class problem. We reveal that these gains increase when less expert data is available, and uncover several interesting properties through further studies. We demonstrate our findings on CSAW-S, a new dataset that we introduce here, and confirm them on two public datasets.
|
|
