[Mittagsseminar TI] Fwd: Einladung Verteidigung Bachelorarbeit

[László Kozma <laszlo.kozma@fu-berlin.de>]

Today in the noon seminar.

Best,
Laszlo

-------- Forwarded Message --------
Subject: [i-prof] Einladung Verteidigung Bachelorarbeit
Date: Wed, 1 Mar 2023 13:32:45 +0100
From: Arman Durmus <armad00@zedat.fu-berlin.de>
To: maria.koekenhoff@fu-berlin.de
CC: i-profs@inf.fu-berlin.de, i-wimis@inf.fu-berlin.de, 
i-studi@inf.fu-berlin.de

Sehr geehrte Damen und Herren,

ich lade Sie hiermit zur Verteidigung meiner Bachelorarbeit mit
dem Titel ""An Algorithm for Hierarchical Clustering - Experiments and
Analysis" ein. Der Vortrag wird am 02.03.2023 um 12:00 Uhr in der
Takustraße 9 Raum 055 auf Englisch stattfinden.

Erstgutachter: Prof. Dr. László Kozma
Zweitgutachter: Prof. Dr. Helmut Alt
Betreuer: Prof. Dr. László Kozma

Viele Grüße,
Arman Durmus

***

Dear Sir or Madam,

I hereby invite you to the defense of my bachelor thesis titled "An
Algorithm for Hierarchical Clustering - Experiments and Analysis". The
presentation will take place on 02.03.2023 at 12:00 am in Takustraße 9
room 055 in english.

First examiner: Prof. Dr. László Kozma
Second examiner: Prof. Dr. Helmut Alt
Supervisor: Prof. Dr. László Kozma

Best regards,
Arman Durmus

***
Zusammenfassung(Abstract):

We present and analyze a new algorithm idea to hierarchically cluster data,
and show that it performs well compared to other established hierarchical
clustering methods when evaluated based on Dasgupta’s cost function. We
prove that in one iteration, our algorithm finds the optimal hierarchical
clustering tree for the given order of the points, when limited to
hierarchical clustering trees the leaves of which have the same ordering
as said points.

Additionally, we show that the cost of the found tree can only improve or
remain the same with each iteration. Thus, the presented algorithm can be
used for hierarchical clustering tasks, or on the results of other
hierarchical clustering algorithms to improve those, and
leaves room for theoretical improvements and experimentation.