BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Chao Tian (Texas A&M University) DTSTART:20240228T160000Z DTEND:20240228T171500Z DTSTAMP:20260423T021450Z UID:AAIT/29 DESCRIPTION:Title: Co mputer-Aided Investigation of Information-Theoretic Limits: An Overview.\nby Chao Tian (Texas A&M University) as part of Seminar on Algorithmic Aspects of Information Theory\n\n\nAbstract\nThe linear programming (LP) f ormulation of information measures provides a solid mathematical framework to identify the fundamental limits of information systems computationally . A critical issue of this approach is however its high computational comp lexity. To reduce the computation burden of this approach\, we can utilize the symmetry structure in such systems. The strength of the symmetry-redu ced approach is illustrated in several well-known difficult problems\, suc h as regenerating codes\, coded caching\, and private information retrieva l\, which provides new and non-trivial outer bounds. In addition to rate b ounds\, more in-depth studies can be conducted on the joint entropy struct ure of these computed bounds\, which often lead to reverse-engineered nove l code constructions and further allow disproving linear code achievabilit y. Finally\, we discuss two new directions: the first is to allow the util ization of non-Shannon-type inequalities in the computational approach\, a nd the second is to convert the original LP into a sequence of smaller LPs \, both of which appear to be awaiting certain suitable machine-learning t echniques.\n LOCATION:https://researchseminars.org/talk/AAIT/29/ END:VEVENT END:VCALENDAR