Comprehensive coverage

Black belt, pink future

The resume of Kira Radinsky, a doctoral student in the Faculty of Computer Science, is impressive even if you ignore the fact that she is only 25 years old * the future is ahead of her - and it is the subject of her research

Kira Radinsky, PhD student at the Technion. Photo: Technion spokespeople
Kira Radinsky, PhD student at the Technion. Photo: Technion spokespeople

Recently we were informed in the world news (And we even reported it on the science website the other day) because researchers from "Microsoft" in the USA and the Technion in Haifa are working on developing software that can help predict natural disasters, increasing violence, epidemics, social unrest and other events that result in mass death. Tests conducted on the software they developed showed that it was possible to identify signals predicting disasters 70-90% of the time. This is not a new technique, but the innovation in the research of Kira Radinsky of the Technion and Eric Horowitz of the "Microsoft" research group in Redmond, which is the first time that so many sources of information were used to improve the system and make it flexible.

The software was able to detect cholera outbreaks in Angola by analyzing news about the drought that hit the country in 2006 and storms in Africa that occurred in early 2007. This after in other cases, after disasters of this scale, a cholera epidemic occurred, both in the history of Angola itself and in other regions of the earth.

Kira Radinsky did her doctorate in the Faculty of Computer Science at the Technion under the supervision of Professor Shaul Markovich. Kira's story is a story of excellence, without a scratch or a rustle and all the way through.
At the age of five she started studying karate, and at the age of fifteen - computer science at the Technion. Since her release from the Intelligence Corps, she managed to establish a business in the Internet field, work in a start-up called "WEBSHAKES" and go on to a "little more" known company - "Microsoft". In between, she completed her bachelor's degree (as part of the Technion's excellent program), started a master's degree and won a series of awards, including the outstanding project award (on behalf of the faculty), the technological innovation award (on behalf of the Technion) and the Anita Borg award (on behalf of Google).

Then she switched to a direct path to a doctorate in the Faculty of Computer Science, under the guidance of Professor Shaul Markovich. Her life at the Technion was busy - at the same time as studying, she served as an intern (excellent) in the faculty, worked at "Microsoft" and was active in the Technion karate team (she has a black belt). "I believe there is an essential connection between karate and research," she says. "In both cases, you must have courage and faith to get up and try again - even if you don't know what the result will be. When you believe in yourself, you have no reason to give up."

As part of her master's degree, Kira Radinsky dealt in areas that are all related to temporal learning and artificial intelligence. Initially, she developed a methodology for predicting future events based on Internet queries - she looked for jumps ("peaks") in the appearance of specific terms in these queries, and based on these she identified causal patterns between different events. Based on those links - in retrospect - she tried - and succeeded - to predict the distribution of future queries. For example, based on publications by the "Apple" company about a new product, the software predicted a flood of queries about Apple shares. "In this case there is an obvious causal relationship, but the fact that it is a statistical mapping allows us to come up with less predictable relationships as well," she says. "For example, we were able to predict the future popularity of the words 'oil' and 'stocks' the day after a 'peak' of queries that included the combination 'decline in the value of the dollar' (note: the original search was done in English)." Equally interesting is its success, using the same algorithms, in predicting hurricanes based on a spike in queries on this topic on the web.

From predicting based on the frequency of queries, Radinsky moved to predicting future events based on computer analysis of news archives. "We analyzed the news published in the New York Times from 1800 to the present day, trying to infer causal relationships. The algorithms we developed make it possible to identify patterns of causality from successive events in the past, and to predict future events based on that causality. This is how we were able, for example, to predict an increase in the price of oil following hurricanes."

Said technology has also been applied to encyclopedic databases, which work differently than news archives. There, too, significant learning was achieved and causal patterns were identified - identification that increases the array of predictive tools available to the user. The algorithms developed by Radinsky also allow the computer to learn cause words (caused, for example) which it can use in the future, to identify additional causal relationships.

"The computer is not smarter than the person, but it is able to handle a larger amount of data and find unexpected links. Since it is a process of learning (from generalizations), which relies on a semantic analysis of events, we manage to identify causal links that we would not have found through human intuition alone," says Kira.

Not long ago, she moved from Haifa to Zichron Ya'akov - on the way between her studies at the Technion and her job at Microsoft, Herzliya. As part of her doctorate, she continued to work in this area of ​​predicting future events based on semantic analysis and time-dependent learning, and did not give up karate and her other hobbies - tennis, reading and writing poetry, salsa dancing, running and training in the gym.
Pictured: Kira Radinsky. Photo: Technion spokespeople

14 תגובות

  1. Every algorithm is based on prejudices and statistics and there is no way to predict the future! In every human culture every generation worships pretenders who claim to predict the future.
    The girl is very good at network marketing, public relations, promotion, communication, sales and a good student. I know the people who work and how the industry works.
    What professional material has she published that has been peer-reviewed?
    Academic research is not industry research.

  2. The girl may be terribly talented, I can't assess because I don't know her, but this is true for both the other commenters and the Technion's public relations. And while she has amazing PR, that doesn't say anything about her scientific abilities either positively or negatively.

    But there is usually a simple test to assess the quality of a scientist's work at the beginning of his career, and that is the amount of his scientific articles and the place of their publication. From a search in the databases, the doctor has a total of one article published in a peer-reviewed journal, a mediocre product for a PhD graduate, although some graduate with no articles at all, but the majority graduate with 2-4 articles in the research field. When the best put out more articles. It is indeed a problematic measure, but it is a very acceptable measure to judge a researcher's abilities, especially in his initial stages.

    And in addition, the same article she published is not directly related to the subject in question here. So I can't claim anything about her abilities or originality (without cynicism) but in the standard terms in which a young researcher is measured she is not an academic elevation.

    Regarding the Technion's public relations, why didn't they provide a reference to any article? Maybe we could read her work ourselves and appreciate it?

  3. I have now seen the article that was broadcast tonight in the Friday studio.
    Don't know how successful her system is, but the whole thing is starting to remind me of Harry Seldon and Psychohistory from Asimov's Institution series.
    It seems that the future is already here, and more and more governments and entities will start using such techniques in the coming decades. There is no reason why they should not do this, because the computing capacity is already here.
    The good uses that can be made of such systems are clear, but it is also clear that there will be bad uses by negative entities, whether they are governments or powerful economic entities.

    Which is sure to be fascinating.

  4. How hard it is to achieve such achievements and how easy it is not to be impressed by them...
    I don't know about you, this young lady's abilities are enviable

    And maybe not, this girl is very talented

  5. Shmulik.

    Kira or someone close to her published two articles at the same time, one in YNET (as I think) and one here. The article in YNET (as I thought) talked about the forecast of an outbreak of political shocks in Syria. I referred to these two articles as a whole, therefore it is legitimate to write here about the prediction about Syria even though this matter was not mentioned here.

    When I was criticized I made a time estimate when she prophesied about Syria. According to an almost certain estimate, based on her resume, the computerized system in question was ready two years ago at the earliest. With a little less certainty, the system was only ready in the last year (and the reason for publication *today* is that the system was completed and finally tested in the last few months).

    With the easy assumption that the system was ready two years ago, the prediction about Syria was made in the last two years. To remind you, already two years ago we were after the uprisings in Tunisia, Libya and Egypt, they called it the "Arab Spring" in the newspapers. This means that predicting an uprising in Syria like this could be done according to common sense, without any computerized system.

    The girl is talented, maybe very talented, but the system she built is not amazing (that's why the advertising is exaggerated).

  6. It seems that the girl is talented and I hope she continues to succeed.

    Another matter is the writing style of the Technion's public relations office/spokesman, which seems to have been taken from another century and another regime.

  7. Safkan,
    No prediction is expected and you said (not the article) that she predicted some prediction about Syria, without specifying the time when she did it, which makes all your words petty. What if she did it a year before everyone else? It is clear to everyone that the article, as nice as it is, is not deep enough to describe the capabilities of the described software, while here everyone jumped as if Kira was the enemy of the people. The article is about Kira and her achievements and they deserve all the respect

  8. Shmulik

    I don't see anything remarkable about the system she built, at least according to the description given here.
    Experimental systems are constantly being built as if a practical product does not come out of the decisive multiple. Their benefit, if any, is that they are part of ant work that is an infrastructure or background for successful systems.

    My cynicism about the prediction of an uprising in Syria is because they made this claim in another publication of the same system, if I'm not mistaken. After all, a prediction about an uprising in Syria is an expected prediction without any help from a computerized system. I can predict that Jordan will also have such an uprising, sooner or later, again without any computerized system.

  9. Kira also predicted an uprising in Syria, while the Primitives predicted an uprising there without relying on any complicated algorithm. Very strange how the primitives succeeded in divination.

Leave a Reply

Email will not be published. Required fields are marked *

This site uses Akismat to prevent spam messages. Click here to learn how your response data is processed.