Showing posts with label IBM. Show all posts
Showing posts with label IBM. Show all posts

Friday, October 30, 2015

IBM partners with the Jacobs School to offer free workshops on IBM Bluemix/Watson


Representatives from IBM provided a free student and faculty workshop with the von Liebig Entrepreneurism Center at the Jacobs School of Engineering on Wednesday, Oct. 21, to introduce how IBM’s cloud Bluemix/Watson Platform can be utilized as a service. The first workshop taught researchers how to use IBM’s Watson product on the Bluemix platform for a host of cognitive computing applications, such as robotics, lasers, imaging and medical diagnostics.   
As part of IBM’s Academic Initiative Program, students and faculty at UC San Diego will be able to use this cloud platform for free. Recently, the von Liebig Center was named as an authorized incubator for the IBM Global Entrepreneur Program.The von Liebig Center plans to continue working with IBM in offering access to more workshops and providing additional support to those who are interested in pursuing commercialization of projects ideas with the use of the Bluemix platform.

At the first workshop, Armen Pishdotchian, Technical Enablement Specialist of IBM Watson Academic Practice, shared that one of the early tests of Watson was used for the Jeopardy game show with host Alex Trebek. He explained that Jeopardy was an easy application and the academic uses are much more advanced. UCSD Faculty expressed interest in using Watson for time series analysis in digital healthcare data, sensors and diagnostics, while students expressed interest in bioinformatics, sensors and robotics. During his introduction to cognitive computing, Dr. Ali Arsanjani, IBM Distinguished Engineer, said that we will eventually learn to trust robots like we have learned to trust our computers.  

The 4 L’s common to Cognitive Computing are: 1. Language (Watson has to know that a wise man is different from a wise guy.)  2. Levels (Potential responses are scored and ranked.) 3. Learning (Watson must be able to recognize patterns in complex data and images.) 4. Limbs (Watson must be capable of understanding robotics and the Internet of Things).


Watson works in 4 stages: 1. Question Analysis (similar to a keyword search). 2. Hypothesis Generation (candidate answers are listed).  3. Scoring (Linguists score the validity of each potential answer).  4. Ranking (Final merge based on the highest score.)  
Both speakers emphasized the versatility of Watson. Watson has been used to map personality traits with the type of work that people should be doing, as well as used to answer questions about insurance, mortgages or any other area of expertise. In the medical field, Watson was used to conduct research on amino acids for cancer treatment as well as a diagnostic tool for physicians to analyze Asperger’s or Down-Syndrome.

Matias Rico, Business Development Manager, IBM Academic Practice holds office hours at the von Liebig Center every other week on Mondays from 11:00 am to 3:00 pm

For more information, please contact Lori Deaton at the von Liebig Center (ldeaton@ucsd.edu).

Monday, August 11, 2014

Lead architect of IBM's 1 million neuron neuromorphic chip is a Jacobs School alum

Huge congratulations to Jacobs School alumnus Dharmendra Modha, the principal investigator on the DARPA SyNAPSE project, which announced last week that it had built a one-million-cores neuromorphic chip and published its results in the journal Science.

The chip, called True North, was developed at IBM Research, where Modha leads the company's Cognitive Computing Group. It's "a synaptic supercomputer in your palm," Modha writes on his blog.
It also consumes dramatically less power than a supercomputer--70 milliwatts, or what it takes to power a hearing aid, according to The New York Times.

Also on the blog, Modha also writes:
The architecture can solve a wide class of problems from vision, audition, and multi–sensory fusion, and has the potential to revolutionize the computer industry by integrating brain–like capability into devices where computation is constrained by power and speed. These systems can efficiently process high–dimensional, noisy sensory data in real time, while consuming orders of magnitude less power than conventional computer architectures.
Modha also wrote a related article on the IBM Research website:

Introducing a Brain-inspired Computer 
TrueNorth's neurons to revolutionize system architecture

Modha earned a Ph.D. in electrical engineering in 1996, under now-retired professor Elias Masry and Robert Hecht-Nielsen, now an adjunct professor in ECE.
When we asked him back in 2009 what he had learned here at the Jacobs School, he said:
I owe a great debt to UCSD, and to my many wonderful professors from the Jacobs School, especially Elias Masry and Robert Hecht-Nielsen, for their priceless gift of education," says Modha, who recently returned to the Jacobs School for a visit. "Fundamental training persists. Professors change lives when they teach thinking, when they teach students how to approach any problem - how to rigorously break down any problem and deal with it," says Modha.
"Now, as a proud UC San Diego alumnus, I am staying connected and working with our university to raise the visibility of the accomplished students, faculty and alumni who add value to our degrees every day."