Skip to main content

IBM R&D Labs in Israel

image: IBM and Haifa

Can machines learn? You bet

Analytics experts flock to Haifa to learn forward-looking solutions in machine learning

Shai Fine at the Machine Learning Seminar
Research scientists are bringing smarts to a huge variety of computerized systems, from medical diagnoses based on clinical genomic data, to the discovery of leaks in water pipes, and new ways to predict which customers may leave their service provider. All this is being done using a special area of analytics known as machine learning.

A branch of artificial intelligence, machine learning lets computers process and learn from vast amounts of data through algorithms. These algorithms can recognize complex patterns and make intelligent decisions based on them. Recently, the IBM Research Lab in Haifa hosted a full day seminar, bringing a record number of participants to hear what's going on at the forefront of machine learning.

"Machine learning and analytics are becoming more critical all the time and we're getting a peak into some of the most advanced research being done in industry and academia," explained Shai Fine, senior manager of analytics at IBM Research – Haifa and one of the seminar organizers. "The seminar brought together some great minds for networking and has also proved itself as a valuable event for meeting potential candidates and partnering on projects with various academic groups."

Promoting collaborations

Both the advanced research in IBM and activities such as this one have established Fine's department as leaders in the Israeli machine learning community. Just last year, the team hosted ICML and COLT, the leading international conferences for machine learning. This leadership position has paved the way for new opportunities and collaborations such as the open collaborative research being done in partnership with Tel Aviv University, the distribution of faculty awards in leading Israeli universities, and membership in high level European Union consortiums.

Machine Learning Seminar in Haifa, June, 2011
The program for the Machine Learning seminar included an impressive lineup of speakers from IBM, HP Labs, Google, Yahoo!, the Technion, Tel Aviv University, and Bar Ilan University. Especially interesting was the trend towards online learning, where the system uses constant adaptation to build a model for classification and prediction, and then continues to enhance the model while data flows in.

"There is a lot of state-of-the-art working being done in the areas of recycling data, classifying information from web applications, or optimizing email and other data," continued Fine. Although previous years' seminars focused more on bioinformatics, many of these topics were highlighted as central issues in the Clinical Genomics Workshop hosted by IBM Research – Haifa the week beforehand.

According to Michal Rosen-Zvi, organizer of the Clinical Genomics Workshop and one of the managers in Fine's department, many people come to the seminars specifically to discuss joint work with IBM. This year's workshop focused on using machine learning in the healthcare domain and presented such topics as whole genome analysis methods and next generation sequencing, an inexpensive method that is becoming popular for use in finding genetic mutations while researching various types of cancer.

Promoting collaborations

Rosen-Zvi's team is currently partnering with leading academic institutions to develop new solutions for genetic research – all based on advanced analytics and machine learning techniques. In one project with Tel Aviv University, they developed an innovative method to share biological data and continue to identify statistical relationships, without reducing the significance of the experimental results. In other work, they are partnering with teams across Europe to understand the role of genetics in the case of essential hypertension. Here, data from thousands of patients and healthy people is accumulated to compare their genetic profile and disease status.

"Many areas of information technology are desperate for ways to deal with the flood of information coming from so many different sources," noted Fine. "When it comes to machine learning, we couldn't ask for anything better. As the volume of information grows, we continue to find deeper insight and discover new opportunities where IBM can help make systems smarter."


Leadership seminars

Machine learning and Jeopardy!

Watson, the IBM DeepQA computing system that bested its human challengers on Jeopardy! uses machine learning algorithms to rank and estimate confidence in possible answers. Watson uses sentence context, for example, to determine how words with multiple meanings are being used. Meet the Haifa team that worked on Watson.

Your IBM expert

Shai Fine, Senior Manager, Analytics, IBM Research

Shai Fine: Senior Manager, Analytics, IBM Research.

Michal Rosen-Zvi: Manager, Machine Learning and Data Mining Group, IBM Research

Michal Rosen-Zvi: Manager, Machine Learning and Data Mining Group, IBM Research