Phymaths podcast # 75 || Dr. Christian Ferko

About this episode

DescriptionDr. Christian Ferko is a theoretical physicist at Northeastern University and IAIFI. He primarily works on sigma models and machine learning (especially on the correspondence between neural networks and QFTs). His work can be found at the link below.Some links1) His works: https://inspirehep.net/authors/18380252) Link to the high-energy physics resources page on my website: https://hassaansaleem.com/graduate-level-theoretical-physics-resources-in-progress/Contact InfoMy website: hassaansaleem.comFollow on Instagram: https://www.instagram.com/hassaan.3142/Join the WhatsApp channel: https://whatsapp.com/channel/0029Va9DkjmAzNbs7IlGuv1XFollow on Twitter: https://twitter.com/Hassaan_PHYEmail Queries: [email protected]Time stamps(0:00) Intro of the guest(0:20) Neural Networks/QFT (NN/QFT) correspondence(1:54) Getting gauge theory(4:07) Adding spinors(6:45) RG scale in Neural Networks (NNs)(9:20) Activation functions(14:40) Other architectures(16:01) Adding interactions(18:05) Anomalies in NNs(23:50) T duality in NNs(30:16) Discrete θ angles in NNs(33:21) Entanglement in NNs(35:41) NNs and Holography(39:45) NNs and generalized symmetries(41:36) TTbar deformation(44:37) TTbar and supersymmetry(47:42) TTbar and Modmax theory(50:48) TTbar and holography(56:54) Outreach(1:05:50) Psychological aspects of research(1:12:00) Choosing problems(1:17:04) Building collaborations(1:24:20) Using AI for research(1:30:08) Choosing Ph.D. students(1:34:22) (People's question) Non-linearity in ML(1:37:43) (People's question) Is AI curious?(1:39:45) (People's questions) Practices for using AI