2 papers
cs.LG2025
Quantifying the Ease of Reproducing Training Data in Unconditional Diffusion Models
Masaya Hasegawa, Koji Yasuda
Diffusion models, which have been advancing rapidly in recent years, may generate samples that closely resemble the training data. This phenomenon, known as memorization, may lead…
hep-th1997
Non-Abelian Gauge Configuration with a Magnetic Field Concentrated at a Point
Minoru Hirayama, Takeshi Hamada, Masafumi Hasegawa
A specific SU(2) gauge configuration yielding a magnetic field concentrated at a point is investigated. Its relation to the Aharonov-Bohm gauge potential and its cohomological mean…