How to deliver on Machine Learning projects • Hacker News Board

Re: How to deliver on Machine Learning projects

#21

Post by yonkshi » Fri, Oct 05, 2018, 3:45 PM UTC

chippy wrote: post #15

vjsc wrote: post #5

So we had this idea of a new feature for our product. The only way to quickly do it was to somehow implement a machine learning algo and that would give us the result that we wanted. Viola!! It seemed simple. Now our company doesn't have any machine learning expert or a data science genius. Going for hiring one would take time. Taking someone up on contract would be very expensive (our CEO wasn't ready to shell out t…

I've been talking to academic neural net / ML experts in computer vision and OCR / NLP and the thing they try to stress is that for almost all cases an algorithmic approach works better.

I don't think most ML experts would agree with that, a big reason DL became popular are the huge improvements they brought to CV and NLP fields.

In many ways, traditional approaches were harder because you need huge amount of domain expertise in CV & NLP, whereas a ML expert can solve simple CV problems with almost no domain knowledge.

Now, a lot of the business data, especially time series data, I agree that an algorithm/heuristic approach is easier and more robust. E.g. recommendation systems.

Re: How to deliver on Machine Learning projects

#22

Post by zwieback » Fri, Oct 05, 2018, 3:49 PM UTC

superflyguy wrote: post #12

vjsc wrote: post #5

So we had this idea of a new feature for our product. The only way to quickly do it was to somehow implement a machine learning algo and that would give us the result that we wanted. Viola!! It seemed simple. Now our company doesn't have any machine learning expert or a data science genius. Going for hiring one would take time. Taking someone up on contract would be very expensive (our CEO wasn't ready to shell out t…

"The only way to quickly do it was to somehow implement a machine learning algo and that would give us the result that we wanted. Viola!!" Designing the perfect viola using machine learning doesn't sound like it's something for beginners.

Just take a violin and scale by a factor of 1.2 or so.

Re: How to deliver on Machine Learning projects

#23

Post by pletnes » Fri, Oct 05, 2018, 4:06 PM UTC

tedivm wrote: post #18

hellogoodbyeeee wrote: post #8

Earlier quoted context omitted.

They gave you two weeks to become a data scientist and implement a working solution? That's nuts. I'm still pretty early career, but I have done data science work for about four years now and I wouldve quoted at least two months to figure out data, clean it, feature engineer, run models, compare results, and then deliver the best performing solution.

And they didn't even have data!

No data cleaning required. That’s often 80% of a project. So 2 months -> 2 weeks makes sense now!

Re: How to deliver on Machine Learning projects

#24

Post by aaronblohowiak » Fri, Oct 05, 2018, 4:08 PM UTC

vjsc wrote: post #5

So we had this idea of a new feature for our product. The only way to quickly do it was to somehow implement a machine learning algo and that would give us the result that we wanted. Viola!! It seemed simple. Now our company doesn't have any machine learning expert or a data science genius. Going for hiring one would take time. Taking someone up on contract would be very expensive (our CEO wasn't ready to shell out t…

While not ML it still would have been considered a form of AI back in the day — “expert systems” they used to call it :)

Re: How to deliver on Machine Learning projects

#25

Post by ende » Fri, Oct 05, 2018, 4:28 PM UTC

reureu wrote: post #20

I love that "Data Scientist" has become such an inflated and meaningless title that now we have "Machine Learning Engineer".

Well, “Data Scientist” has been appropriated by the overflow of PhD’s w/o any actual stats or computational backgrounds and few academia prospects, so I guess you need to create new job titles for thise who are going to do the actual work.

Re: How to deliver on Machine Learning projects

#26

Post by jimcsharp » Fri, Oct 05, 2018, 4:39 PM UTC

dragandj wrote: post #9

vjsc wrote: post #5

So we had this idea of a new feature for our product. The only way to quickly do it was to somehow implement a machine learning algo and that would give us the result that we wanted. Viola!! It seemed simple. Now our company doesn't have any machine learning expert or a data science genius. Going for hiring one would take time. Taking someone up on contract would be very expensive (our CEO wasn't ready to shell out t…

> The only way to quickly do it was to somehow implement a machine learning algo and that would give us the result that we wanted. Since no one of you had any experience with ML, how did you know that a ML algo (which one?), implemented "somehow" would give you the results you wanted? (Not a cynical comment; I am really interested in hearing about this).

Not OP, but went through a similar situation and the feature was 'alert us about unintuitive correlations in our data so we can invent new KPIs'

Re: How to deliver on Machine Learning projects

#27

Post by pedrosorio » Fri, Oct 05, 2018, 4:58 PM UTC

yonkshi wrote: post #21

chippy wrote: post #15

Earlier quoted context omitted.

I've been talking to academic neural net / ML experts in computer vision and OCR / NLP and the thing they try to stress is that for almost all cases an algorithmic approach works better.

I don't think most ML experts would agree with that, a big reason DL became popular are the huge improvements they brought to CV and NLP fields. In many ways, traditional approaches were harder because you need huge amount of domain expertise in CV & NLP, whereas a ML expert can solve simple CV problems with almost no domain knowledge. Now, a lot of the business data, especially time series data, I agree that an algo…

"traditional approaches" in CV & NLP were also ML (a quick reminder that machine learning existed long before the deep learning hype).

Not sure what the parent meant by "algorithmic approach" though.

Re: How to deliver on Machine Learning projects

#28

Post by idontpost » Fri, Oct 05, 2018, 4:59 PM UTC

zwieback wrote: post #22

superflyguy wrote: post #12

Earlier quoted context omitted.

"The only way to quickly do it was to somehow implement a machine learning algo and that would give us the result that we wanted. Viola!!" Designing the perfect viola using machine learning doesn't sound like it's something for beginners.

Just take a violin and scale by a factor of 1.2 or so.

I feel like the devil is in the at "or so" part.

Re: How to deliver on Machine Learning projects

#29

Post by sidr » Fri, Oct 05, 2018, 5:03 PM UTC

vjsc wrote: post #5

So we had this idea of a new feature for our product. The only way to quickly do it was to somehow implement a machine learning algo and that would give us the result that we wanted. Viola!! It seemed simple. Now our company doesn't have any machine learning expert or a data science genius. Going for hiring one would take time. Taking someone up on contract would be very expensive (our CEO wasn't ready to shell out t…

I like this resource a lot for new data scientists: https://developers.google.com/machine-learning/guides/rules-... . Rule #1 seems pertinent to your situation.

Re: How to deliver on Machine Learning projects

#30

Post by reureu » Fri, Oct 05, 2018, 5:04 PM UTC

ende wrote: post #25

reureu wrote: post #20

I love that "Data Scientist" has become such an inflated and meaningless title that now we have "Machine Learning Engineer".

Well, “Data Scientist” has been appropriated by the overflow of PhD’s w/o any actual stats or computational backgrounds and few academia prospects, so I guess you need to create new job titles for thise who are going to do the actual work.

I totally agree, and wasn't arguing that a new title wasn't necessary. And I'm ok with my downvotes for that comment :)

It's just funny that "Data Scientist" seemed to be originally branded as the more technical/engineer-y version of a data analyst. Now I get recruiters contacting me for "Data Scientist" positions that entirely revolves around SQL and excel, and nobody in the Bay Area hires "Data Analysts" anymore.

Alright, guess it's time to update my LinkedIn and resume to adjust for this inflation? Maybe I should jump up a few inflation levels and just become a "Deep Learning Engineer."