IBM Outlook and Collaboration Event
March 14, 2010
Organized by TAU/CS and IBM Haifa Research Lab
| Abstracts |
IBM's wire-speed processor as a research opportunity
Dr. Amit Golander, IBM Research – Haifa
Tel-Aviv University has been selected as one of the first three universities worldwide to perform research on IBM's newest technology: the Wire Speed Processor. The processor is a hybrid machine that employs massive multithreading capabilities, integrated IO, and unique special-purpose accelerators such as Compression, Crypto, regular expressions, XML, and TCP.
In this talk I describe the research opportunity and why it can facilitate the development of faster algorithms in many fields of computer science such as Networking, Cryptography, Language, Bioinformatics, SOA, and Systems.
You are invited to an additional seminar lecture presenting the processor, which will be held on March 8, 15:00, in room 11 of the Electronic Engineering classrooms building:
http://www.eng.tau.ac.il/index.php?option=com_jevents&task=icalrepeat.detail&evid=27&Itemid=292&year=2010&month=03&day=08&uid=8b9cead2e2279cf6686fc686aedeabc3&catids=176|178|177|181|182|179|180|183
Emerging challenges of Cloud Computing
Dalit Naor, IBM Research – Haifa
Cloud Computing, probably the most fascinating new paradigm in computer systems today, is designed to deliver IT as a service in a pay-per-use model over the internet. It is presenting the computer science research community with a wealth of new research topics related to scalability, security, cost models, and service levels. IBM Research – Haifa is conducting active research in cloud technologies, for both the compute and storage clouds. An interesting area of investigation involves federation across clouds. For the compute cloud, the aim is to develop technologies for the federation of clouds, allowing live migration of compute workloads across disparate clouds, thus fulfilling the premise of a single cloud with infinite computing capacity. For the storage cloud, the objective is to avoid the data lock-in that is a main barrier today for storage cloud adoption, by ensuring data migration and interoperability across different cloud providers. In this talk I summarize some of the technical challenges in cloud federation today, and describe various approaches to eliminating technical barriers.
Crowd computing: are we close to assessing the power of the crowd?
Ohad Greenshpan, IBM Research – Haifa
With the realization of Web 2.0, people have begun to actively contribute content via various means, collaborate via emerging communities, and build bodies of knowledge.
There have been a number of successful attempts to harness people to solve difficult problems. The natural skills of humans in the fields of vision and speech, along with the general accumulation of knowledge in all of our minds, are what make applications that use them so powerful. The paradigm that deals with harnessing humans to solve computationally troublesome problems is called Crowd Computing. These Crowd Computing applications assume participation of the masses, where each participant solves a small portion of the problem, and together the crowd solves the entire problem. Since Crowd Computing applications involve humans to assist in computations, they might change the way new algorithms are developed for hard problems in various CS fields such as machine learning, computer vision, speech recognition, and NLP.
In my lecture I review some fascinating examples of Crowd Computing applications and discuss the challenges that are expected to arise in this field in the future.
Machine Learning in the Real World: From Wafer Manufacturing to Genomic Data Analysis
Noam Slonim, IBM Research – Haifa
The explosion of available information in various domains is creating new opportunities as well as new challenges for the machine learning community. In the Analytics group at HRL, we have approximately 30 researches, working on real world problems that involve experimental and theoretical research in machine learning and constraint satisfaction problems. In this presentation I will briefly review some of the exciting projects we are involved in, ranging from improved wafer manufacturing to genomic data analysis and support for clinical treatment.
Panel: What are the next "hot" fields and research topics?
Below are some questions that will be raised at the panel discussion:
- Academic outlook: What are the next "hot" fields in academic research? Are they interdisciplinary or pure computer science?
- Industrial and academic research: What are the differences between the expected "hot" research topics and the fields that will be the focus of the industrial research? Is there an overlap? How can we explain the differences and narrow the gaps?
- Research versatility: How can we change our research topics? Do we benefit from the change? Do we have a lower chance of succeeding in the new fields? In which types of research (academic or industrial) do we need to be more flexible in our research interests?
- Collaboration with IBM: What are the benefits and how can we promote it?
