pesenti

27 ms·

211.
by pesenti 10y ago
I know IBM and Watson often get a bad rap here. But IBM made the same exact move almost 3 years ago with the creation of the Watson unit. There is way too much PR around Watson, but IBM should be credited for having called the current round
212.
Artificial intelligence needs a reboot, say experts. (2011)(technologyreview.com)
1 points by pesenti 10y ago|0 comments
213.
Visualization and analysis of 62 000 people clicking inside a square(boltkey.cz)
2 points by pesenti 10y ago|0 comments
214.
by pesenti 10y ago
No you looked at the wrong figure. It's figure 7 (Switchboard) and the WER for humans is 4%.
215.
by pesenti 10y ago
The human error rate on this task is 4% ( https://pdfs.semanticscholar.org/387e/7349b8e31e316c2a738060... ). So you are basically saying that telephones are a useless interface... What is amazing is how consistently peop
216.
by pesenti 10y ago
The human error rate on this task is 4% ( https://pdfs.semanticscholar.org/387e/7349b8e31e316c2a738060... ). What is amazing is how consistently people overestimate human performance.
217.
Watson Conversation API(ibm.com)
4 points by pesenti 10y ago|0 comments
218.
Find Pokemon with Watson Visual Recognition(youtube.com)
2 points by pesenti 10y ago|0 comments
219.
Watson Conversation API(ibm.com)
5 points by pesenti 10y ago|0 comments
220.
by pesenti 10y ago
For the mathematically inclined, Terence Tao's explanation provides an easier intuition of the relationship using real-variable methods: https://terrytao.wordpress.com/2010/04/10/the-euler-maclauri...
221.
by pesenti 10y ago
Has any anybody here considered leveraging the on prem offerings of cloud providers, e.g., Azure on prem ( http://www.geekwire.com/2015/microsoft-brings-azure-on-premi... ) or Bluemix local ( http://www.ibm.com
222.
by pesenti 10y ago
Very fair point. Our STT service, even though it is based on a technology with the very best results for public benchmarks, is not as robust to accents at the moment - for the exact reason you mention, we didn't train it on all the dia
223.
by pesenti 10y ago
We are very aware of the Bluemix usability issues and are working with that team to address them. Did you try the new Bluemix by any chance? Thanks for the kind comments on our APIs. We are really trying hard to make them usable.
224.
by pesenti 10y ago
We do have a team dedicated to Machine Translation (the same team that actually introduced the statistical approach - the "IBM Models" - in the 90s) but that team had been focused more on non-European languages. If you search the
225.
by pesenti 10y ago
You should reach out to us or ask questions on the forum. Our experience is that people experiencing very bad accuracy may not have passed the right parameters (e.g., you need to make sure to use the right audio model). We are aware of the
226.
by pesenti 10y ago
No. That's not the issue. We are actually able to shrink wrap it to a point that it has reasonable footprint (a few servers). The problem with factoids are twofold: - It's always a small part of a broader use case (how many querie
227.
by pesenti 10y ago
Sure here are two (you may have to follow some links to go to the actual publications): Speech to text: https://developer.ibm.com/watson/blog/2016/04/28/recent-adva... Emotion: https://de
228.
by pesenti 10y ago
You will indeed be able to do Jeopardy too, and hopefully much more (we found that answering factoid questions isn't the most critical function so we haven't prioritized it, but it's coming). Many of the services that are on
229.
by pesenti 10y ago
If you want to figure out what Watson can do and bypass all the marketing hype, you can just try all the services available at http://ibm.com/watsondevelopercloud . I won't argue that the PR goes often too far, and that
230.
IBM Watson can help find water wasters in California(popsci.com)
1 points by pesenti 10y ago|0 comments
231.
by pesenti 10y ago
For those wondering what technology is being used. It is using this Watson product: http://www.ibm.com/smarterplanet/us/en/ibmwatson/engagement_... . We are in the process of revamping it and making it a
232.
by pesenti 10y ago
It's using this: http://www.ibm.com/smarterplanet/us/en/ibmwatson/engagement_... but if you are smart you can reproduce a lot of the functionality using both http://www.ibm.com/smart
233.
by pesenti 10y ago
It's using this: http://www.ibm.com/smarterplanet/us/en/ibmwatson/engagement_... but if you are smart you can reproduce a lot of the functionality using both http://www.ibm.com/smart
234.
Failure is moving science forward(fivethirtyeight.com)
1 points by pesenti 10y ago|0 comments
235.
by pesenti 11y ago
You didn't read the qualifier: "released to the founders". It means that CMU didn't pursue the tech we invented there. So we were free to go out and do something with it as long as we gave them a small chunk (though nobo
236.
by pesenti 11y ago
You don't. But investors tend to invest a big chunk of their money locally (they don't like to travel for board meetings...)
237.
by pesenti 11y ago
Oh my, never considered Pittsburgh as part of "Appalachia" - I thought that was West Virginia! I guess the US looks at Pittsburgh like Pittsburgh looks at WV...
238.
by pesenti 11y ago
I bootstrapped my startup so I indeed never needed outside money (well rather never really tried given that we started during the 2000-2002 nuclear winter). But if I had to do it again, especially in the current VC environment, I would rais
239.
by pesenti 11y ago
I bootstrapped a company in Pittsburgh in 2000 (the founders all came from CMU - I am from France originally) with a good exit in 2012. When I moved there the food scene was nonexistent and people in the US still perceived Pittsburgh as a b
240.
by pesenti 11y ago
Yes. We have a bunch of cloud services (including visual recognition - demonstrated here) that you can embed in your own application. Try them here: http://ibm.com/watsondevelopercloud

More ›