2026-10-06T18:15:21Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33956172 2026-09-21T17:30:18Z category_1 category_7 category_12 category_15 category_734 category_811 portal_5 item_type_3 month_year_09_2026

<p>Neural network architectures used in the experiments.</p> Jingcong Li (11535825) Yu Liao (3489437) Ximeng Wang (793633) Jie Deng (190312) Biophysics Medicine Cell Biology Neuroscience Biological Sciences not elsewhere classified Mathematical Sciences not elsewhere classified treating region optimization stability statistics calibrate progressively reallocating optimization informed neural networks closed feedback loop also ranks first aggregate error relative partial differential equations greater learning difficulty 0 %, 61 five backbone architectures div >< p resolved collocation points hysics -< b r </ b n </ b co </ b p </ b collocation points </ b poisson equations cognitive learning 6 %, five dimensions five backbones uniform training specific neighborhoods significant problem scientific computing results establish resolve regions paper introduces lowest geometric independent additions free paradigm four methods dimensional benchmarks difficult regions coupled limitations cognitive weights best method although physics 5d poisson <p>Neural network architectures used in the experiments.</p> 2026-09-21T17:30:03Z Dataset Dataset 10.1371/journal.pone.0358646.t002 https://figshare.com/articles/dataset/_p_Neural_network_architectures_used_in_the_experiments_p_/33956172 CC BY 4.0