What I find most exciting here is that they were able to improve performance by combining the three previous steps into one, in other words asking directly for what they wanted rather than trying to trick Hive into…
That is a good origin story! I would contend though that it was the performing of the data analytics on himself (which helped keep it top of mind and helped him to develop an internal model) that was the most helpful.…
I'm a diabetic (type one, since 1988) who has also been doing ML startups for the last 15 years. My HbA1C scores have always been below 6, controlled with a two to four blood test per day and long/short (currently…
Most teams I have seen have either template scripts or boilerplate that generates datasets, and share both the generated data and the scripts via normal ways that people share data and code: disk, S3, github, emailing…
The Tensorflow code mentions "GCUDACC" in several places, and from the surrounding comments it seems to be targeted at OpenCL as well as CUDA. So it seems that this has been at least considered.
It's very hard to find a viable business model in machine learning or AI tools and platforms at the moment: the size of the market is small, and most money is being made by the end users. As a result companies who could…
In my experience, the architecture supports cards with shader model >= 3.0. Occasionally a commit will break the support (eg https://bitbucket.org/eigen/eigen/commits/a19653b8035d8ace53... was required earlier this…
Better, if the processes are on the same machine you could use it to share the data via shared memory or a common memory mapping, to avoid having copies of the data on each end of the pipe.
Once the open source version of Tensorflow releases multi-node support, this would be one way to make it work. There are potential gains from using a GPU for RF training. As for distributing, in my experience for small…
None of the systems include the data load time, but for mldb and the other non-distributed systems, it's only a few seconds. (edit: my grammar is good not)
There are plenty of alternatives out there to Spark ML: here is a survey of RF implementations: https://github.com/szilard/benchm-ml/tree/master/z-other-too... There is a whole other world of non stochastic gradient…
They do provide some very useful pre-trained models, eg the full parameter set for their Inception model.
For profiling of models, almost everything needed is already there. You only need to pass in a StepStatsCollector through the Session::Run() method (I called it RunWithStats() ) and hook it up to the Executor Args by…
It seems that MLDB would be a decent fit for this use-case. You would be able to do pre-processing in the background continuously, and predictions could do a significant amount of work on-demand. Depending upon the size…
(Founder here). Could you describe your use-case? This is an interesting question and I'd love to hear more about what you are thinking of. There are two main parts to most machine learning workloads: training and…
I would still expect it to tend towards the normal distribution across a large set of documents. If you model positive and negative word counts as a binomial distribution, you have the the difference of two samples from…
You are right; this does just shift the bias, which is sometimes all you need (you have a simple algorithm, presumably for a reason). I did misunderstand that you don't have a training set, just a list of positive and…
Two simple things you could do: 1. Insert each negative example six times into your training set (or weight negative examples accordingly, ie use #positive matches - 6 * #negative matches / (2 * positive word count) as…
What I find most exciting here is that they were able to improve performance by combining the three previous steps into one, in other words asking directly for what they wanted rather than trying to trick Hive into…
That is a good origin story! I would contend though that it was the performing of the data analytics on himself (which helped keep it top of mind and helped him to develop an internal model) that was the most helpful.…
I'm a diabetic (type one, since 1988) who has also been doing ML startups for the last 15 years. My HbA1C scores have always been below 6, controlled with a two to four blood test per day and long/short (currently…
Most teams I have seen have either template scripts or boilerplate that generates datasets, and share both the generated data and the scripts via normal ways that people share data and code: disk, S3, github, emailing…
The Tensorflow code mentions "GCUDACC" in several places, and from the surrounding comments it seems to be targeted at OpenCL as well as CUDA. So it seems that this has been at least considered.
It's very hard to find a viable business model in machine learning or AI tools and platforms at the moment: the size of the market is small, and most money is being made by the end users. As a result companies who could…
In my experience, the architecture supports cards with shader model >= 3.0. Occasionally a commit will break the support (eg https://bitbucket.org/eigen/eigen/commits/a19653b8035d8ace53... was required earlier this…
Better, if the processes are on the same machine you could use it to share the data via shared memory or a common memory mapping, to avoid having copies of the data on each end of the pipe.
Once the open source version of Tensorflow releases multi-node support, this would be one way to make it work. There are potential gains from using a GPU for RF training. As for distributing, in my experience for small…
None of the systems include the data load time, but for mldb and the other non-distributed systems, it's only a few seconds. (edit: my grammar is good not)
There are plenty of alternatives out there to Spark ML: here is a survey of RF implementations: https://github.com/szilard/benchm-ml/tree/master/z-other-too... There is a whole other world of non stochastic gradient…
They do provide some very useful pre-trained models, eg the full parameter set for their Inception model.
For profiling of models, almost everything needed is already there. You only need to pass in a StepStatsCollector through the Session::Run() method (I called it RunWithStats() ) and hook it up to the Executor Args by…
It seems that MLDB would be a decent fit for this use-case. You would be able to do pre-processing in the background continuously, and predictions could do a significant amount of work on-demand. Depending upon the size…
(Founder here). Could you describe your use-case? This is an interesting question and I'd love to hear more about what you are thinking of. There are two main parts to most machine learning workloads: training and…
I would still expect it to tend towards the normal distribution across a large set of documents. If you model positive and negative word counts as a binomial distribution, you have the the difference of two samples from…
You are right; this does just shift the bias, which is sometimes all you need (you have a simple algorithm, presumably for a reason). I did misunderstand that you don't have a training set, just a list of positive and…
Two simple things you could do: 1. Insert each negative example six times into your training set (or weight negative examples accordingly, ie use #positive matches - 6 * #negative matches / (2 * positive word count) as…