Skip to main content
    Israel [change]    Terms of use
    Home    Products    Services & solutions    Support & downloads    My account    
IBM Research

When machines learn

IBM Haifa Labs News Center

June 25, 2006

The annual Machine Learning seminar recently held at the IBM Haifa Research Lab presented a number of leading efforts in the areas of inductive and deductive computerized learning.

Attended by over 150 participants representing Israeli universities and industry, the seminar reflected the deep collaboration taking place in the field between the corporate world and academia. The second such seminar in two years, the IBM Haifa event has become a leading forum for Israeli computer scientists who work in the area of machine learning application and research. Half of the seminar lectures described practical and industrial applications of machine learning, while the other talks presented the more theoretical aspects of the discipline being studied in universities.

Michael Kearns from the Computer and Information Science Department at the University of Pennsylvania delivered the keynote address, in which he discussed the use of machine learning for assessing and analyzing equity market microstructures. Kearns presented a summary of recent machine learning approaches to this discipline, and described how a learning-based technique could help measure the impact of a specific trading strategy.

The seminar focused on a number of themes, including applications of machine learning in computerized vision. Gaby Hayun of MobilEye and Yair Weiss of the Hebrew University in Jerusalem described two projects that use machine learning for vision applications.

The seminar also reflected the growing focus of the machine learning world on bioinformatics and healthcare applications. "Until recently, machine learning had been applied extensively in academic bioinformatics projects," noted Shai Fine, manager of the Verification Solutions and Machine Learning Group in IBM Haifa who co-organized the seminar with Yishai Mansour of Tel Aviv University. "We're now beginning to see an increasing number of industrial applications of machine learning in the healthcare and medical spheres," he explained.

Two presentations in the seminar exhibit that growing trend. Vered Aharonson of NexSig described a system that uses machine learning to make early assessments and predictions of neurological illnesses. Michal Rosen-Zvi of the IBM Haifa Labs explained IBM Haifa's involvement in the EUResist initiative. Funded by the European Union, EUResist is an initiative targeted at predicting the response to HIV treatment. IBM researchers use machine learning to compare the performance of various approaches and develop a model to predict drug resistance.

In his remarks at the end of the seminar, Moshe Levinger, a senior manager at the IBM Haifa research lab, described two major challenges currently facing the machine learning community. "Machine learning techniques must be made more accessible to more people," he noted, "if the field is to continue to develop". He also said that a top technology based solely on machine learning has yet to appear.

"Until now, machine learning has played a supporting role, helping to enhance or improve other technologies" he said. "As this discipline matures, I predict that we will see a machine learning 'killer app' in the next few year-maybe based on one of the ideas presented here today."


Document options
Print this page  

    About IBMPrivacyContact