Κύκλος σεμιναρίων

[sandra]

Έχει αναφερθεί αλλά κι αυτό δεν τα έχει όλα.

[The Punisher]

A Design Framework for Event Delivery Systems
Avigdor Gal, Faculty of Industrial Engineering & Management, Technion, Israel
25 Jun 2010 - 15:00
A41, Antoniadou Wing, Main Building

abstract:
Architectures and middleware for event delivery face scalability challenges in providing up-to-date events to meet clients' specifications. The use of proxy middleware is a common practice for increasing scalability in the face of client sophistication. Proxies can aggregate client specifications, taking advantage of similar needs to reduce communication overhead, workload on event sources (e.g., sensors} and network traffic. However, in a setting with thousands of clients and event sources, and constrained resources, proxies cannot always satisfy all client needs. A proxy is interested in maximizing completeness by capturing as many events as possible. However, due to constraints on resources, event delivery may not be fully current, resulting in a delay in delivering events. In many cases, these two dimensions cannot be directly related and we propose a flexible design framework that can suit different changing needs of designers. We start by casting the event delivery trade-off as a bi-objective optimization problem. The conventional definition of a solution to multi-objective problems is a Pareto set and we provide a polynomial-time solution for a restricted version of this trade-off problem. This set of offline optimal solutions serves us in evaluating the design of run-time heuristic solutions, allowing a flexible framework that suits both designs that emphasize completeness and those that emphasize currency. To demonstrate the proposed design framework, we propose a generic online solution and four greedy local policies instantiations of this solution to generate event collection schedules. We analyze the performance of the offline solution and the four policies using both real-world traces and synthetic data and demonstrate how to use the proposed design framework with the four online policies.

[Theofaman]

Algorithmic Barriers from Phase Transitions

Δημήτρης Αχλιόπτας, τμήμα Πληροφορικής, University of California, Santa Cruz
04/11/2010 - 11:30
Αίθουσα Α41, Πτέρυγα Αντωνιάδου, Κεντρικό κτίριο

For many random Constraint Satisfaction Problems, such as
random k-SAT and random graph coloring, we have very sharp estimates
of when solutions exist. For these same problems, though, all known
polynomial-time algorithms can only find solutions in a very small
fraction of the satisfiable regime. We study this phenomenon by
examining how the geometry of the set of solutions evolve as
constraints are added.
We prove in a precise mathematical sense that, for each problem
studied, the barrier faced by algorithms corresponds to a phase
transition in that problem’s solution-space geometry. Roughly
speaking, at some problem-specific critical density, the set of
solutions shatters and goes from being a single giant ball to
exponentially many, well-separated, tiny pieces. All known polynomial-
time algorithms work in the ball regime, but stop as soon as the
shattering occurs. Besides giving a geometric view of the solution
space of random CSPs our results provide novel constructions of one-
way functions.
http://users.soe.ucsc.edu/~optas/

[sandra]

Θα έρθει κανένας αύριο;

[The Punisher]

εγώ

[darkness]

Κι εγώ. Τελευταία στιγμή το είδα...

[Zifnab]

Πώς ήταν?

[The Punisher]

Πάρα πολύ ενδιαφέρουσα η ομιλία. Προσέγγιζε κάποια κλασσικά προβλήματα της Πληροφορικής με τρόπους δανεισμένους απ την Φυσική. Το συναρπαστικό της υπόθεσης είναι ότι, από δύο εντελώς διαφορετικά μονοπάτια, τα όρια (φραγμοί) στα οποιά καταλήγουμε ως αποτέλεσμα είναι ίδια και με τους 2 τρόπους.

Αν ανέβουν οι διαφάνειες, αξίζει να τους ρίξετε μια ματιά.

[Theofaman]

Extracting medical and ancestry information from genomes
Σεραφείμ Μπατζόγλου
14/12/2010 - 13:00
Αίθουσα Α41, Πτέρυγα Αντωνιάδου

http://seminar.cs.aueb.gr/el

[darkness]

Ω αυτό δεν είναι το περυσινό που είχε ακυρωθεί; Ωραία!

[sandra]

Παιδια κανενα νεοτερο για αυριο; Θα ακυρωθει ΠΑΛΙ;

[Watchman]

Αυριο δεν θα γινουν μαθηματα. Η διαλεξη;

[sandra]

Τελικα ναι θα γινει.

[elsupreme]

Έγινε/πώς ήταν/μπορεί να κάνει κανείς περίληψη/ευχαριστώ !

[darkness]

Active Cyber-Thermal Management of Data Centers for Energy Efficiency
Andreas Terzis, Department of Computer Science, Johns Hopkins University
12 Jan 2011 - 12:00
A41, Antoniadou Wing, Main Building

The IT industry is the one of the fastest growing sectors of the U.S. economy in terms of its energy consumption. According to a 2007 EPA report, U.S. data centers alone consumed 61 billion kWh in 2006 — enough energy to power 5.8 million average households. Even under conservative estimates, IT energy consumption is projected to double by 2011. Reducing data center energy consumption is a pressing issue for the entire IT industry now and into the future. In this talk I will argue that dense and real-time environmental monitoring systems are needed to improve the energy efficiency of IT facilities and describe RACNet, a sensor network for monitoring a data center’s environmental conditions. RACNet overcomes the network’s large scale and density and the data center’s harsh RF environment to achieve data yields of 99% or higher over a wide range of network sizes and sampling frequencies. In the second part of the talk I will describe how data collected by RACnet can be used to predict temperature distributions and trends in data centers. Such accurate predictions inform scheduling algorithms that place workload in ways to improve energy efficiency by reducing the duty cycle of the data center’s air conditioning system and thus its energy consumption.
http://seminar.cs.aueb.gr/en/content/ac ... efficiency

[Theofaman]

Θα πάει κανείς?

[darkness]

Θα πάει κανείς?

Αν και το συγκεκριμένο θέμα δεν συγκαταλέγεται και ανάμεσα στα πιο ενδιαφέροντα για μένα, θα είμαι στην σχολή εκείνη την ώρα και απ' όσο θυμάμαι δεν έχω τίποτα άλλο να κάνω, οπότε μάλλον θα πάω.

[enum21]

Cryptography for the Media Revolution
Άγγελος Κιαγιάς, Τμήμα Πληροφορικής και Τηλεπικοινωνιών, Εθνικό και Καποδιστριακό Πανεπιστήμιο Αθηνών
09/03/2011 - 13:00
Αίθουσα Α41, Πτέρυγα Αντωνιάδου, Κεντρικό κτίριο

The transition of production and distribution of today's media to digital form has lead
to a revolution of content production and dissemination. As with every major technological
advance, this media revolution comes with a dark side that has the potential to counteract
its benefits. Indeed, the quality of being digital pairs characteristic qualities such as ease of production
and rapid dissemination with significant downsides such as facilitating misappropriation and
uncontrolled propagation. The repeated failures of standard computer security methods
to solve these problems in practice, motivated a new class of cryptographic constructs
that include broadcast encryption, traitor tracing, fingerprinting codes and others. In this lecture
I will give a small glimpse to this new cryptographic landscape and briefly discuss recent
constructions and relevant security models.

βιογραφικά στοιχεία:
Aggelos Kiayias works in cryptography and computer security.
His research has been funded by a number of agencies including, the United States
DoD, DHS, NIST and NSF including a CAREER award as well as the EU ERC and Marie Curie
programs. In 2011, he served as the program chair of the Cryptographer's Track of
the RSA Conference. He holds a Ph.D. in Computer Science from the City U. of New York that
was funded by a Fulbright fellowship and is a graduate of the University of Athens,
Department of Mathematics.

http://seminar.cs.aueb.gr/el/content/cr ... revolution

[darkness]

Crowdsourcing using Amazon Mechanical Turk: Quality Management and Scalability
Παναγιώτης Ηπειρώτης, Department of Information, Operations, and Management Sciences, New York University

http://seminar.cs.aueb.gr/el/content/crowdsourcing...scalability

Αύριο στη 1 στην Α41.

[Zifnab]

http://seminar.cs.aueb.gr
Μοιάζει κάπως παρατημένο ελπίζω να είναι η ιδέα μου... Αλλά δεν θα έπρεπε να υπάρχει κάποια ενημέρωση για μελλοντικές ομιλίες?...