Yottamine Serves Up Predictive Analytics On-Demand

By Tiffany Trader

February 15, 2013

Predictive modeling is not new, but startup Yottamine Analytics is counting on the convenience and economics of cloud computing to make it a lot more popular. Its cloud-based predictive modeling solution combines the benefits of EC2 spin-up automation and large-scale program parallelism to provide predictive power by the hour for pennies a minute.

Yottamine specializes in software for predictive analytics, specifically what is often referred to as machine learning. The software examines a large number of examples and develops a model that can then look at new data and make conclusions about it.

Take a spam filter, for example. If you were building a spam filter with this kind of software, you would essentially submit to the software, in a specially encoded form, examples of email messages, ones that are spam and ones that are not spam.

The spam filter example is provided by Tim Negris, the vice president of sales and marketing for Yottamine. Negris has been in the data arena for some time; he was formally an executive at Oracle and IBM as well as a number of other startups.

“What the software does is actually quite brilliant,” he says. “It figures out a mathematical formula that given a new example where it doesn’t know whether it’s spam or not, it can predict whether it is or it isn’t.”

This kind of computation is extraordinarily compute-intensive, says Negris. There’s this gigantic matrix of binary values that requires very complex multi-dimensional mathematics. It burns through a lot of cycles.

This kind of machine learning has been around for a while, but it hasn’t made good economic sense because of the very large computational requirement. “Everyone says the same thing,”  says Negris, “‘This is a very promising type of technology, but the computer you would need to have in your datacenter to do it is sufficiently large to where you’d have a difficult time doing it.’ That’s where HPC in the cloud comes into play.”

Yottamine has teamed up with Amazon Web Services to provide machine learning on-demand. “Normally this kind of process requires a lot of trial and error guesswork on the computation side and a lot of hand-work on the deployment side, including setting up the cluster instances,” says Negris. “Configuring this by-hand is not impossible, but takes some time. We automate a lot of those processes so the data scientist can say ‘here’s my data, spin me up a cluster and let’s get a model built.'”

“Using Amazon definitely advances the potential for this kind of machine learning. There’s an interesting affinity between the cost economics of cloud computing and the way in which you use a computer for this kind of thing. The designer spends a couple days figuring out what they need and getting the data ready and then the model is built and you’re off and running until you need to revise the model or build a new model,” is how Negris puts it.

The Yottamine team has been working on finalizing pricing details and lining up customers in preparation for their official launch, which is scheduled for next week. Depending on the algorithm, pricing will be in the range of $10-$50 per node hour. Expect standard linear machine learning algorithms to have a lower price point than more complex Gaussian variants, which can process enormous amounts of information very quickly.

Negris explains their solutions enable very fast runs, meaning a job that would normally take six days with typical open source research software might take only six hours with Yottamine’s software.

“The time compression is considerable,” he says. “You can do a huge model in the space of five hours. For approximately $250, you have a model that two years ago would have required you to own a two-million dollar supercomputer.”

Next >> In the Beginning

Yottamine founder David Huang has been working in machine learning for many years. He started out as a research post doc making algorithms scalable. He finds machine learning very interesting. “It’s not like some problems in biology where data needs are lower and everything can fit into memory but computation needs are very high and take a long time. Machine learning can be both compute and data-intensive if you use a complex model, which is what gives you higher accuracy,” he says.

Huang recalls his own frustrations coming up with big ideas and being told by his supervisor that they were impractical because there wasn’t enough compute power. So three years ago, he started his own company. He sees a huge opportunity for data scientists to be able to predict a wide range of variables for problems that require a huge amount of computing power.

“To solve these problems, we need to be able to use a lot of computing power, and that is complicated. To transform a machine learning algorithm from a single-threaded design into a parallelized computing environment is very difficult,” Huang explains.

To help them overcome the constraints that he once faced, he has worked to bring highly parallel, cloud-based machine learning software to data scientists.

“Our clients are working data scientists who on a daily basis need to do models for a given application; it could be insurance, it could be stock market or finance, it could be digital advertising. The one thing in common is they want to do a model fast and they want to do a highly-accurate model. Such a model requires a lot of computing power, and that’s where the majority of our use cases are in,” he says.

There are many different kinds of machine learning algorithms, and some are more easy to parallelize than others, but most often the best and most desirable are the hardest to parallelize, notes Huang. He points to support vector machines (SVM), which are very robust and have good performance in terms of accuracy, but require a lot of computing power and are not easy to parallelize. “What we’re doing is taking one of the best algorithms out there and parallelizing it so it’s scalable and provides robust performance for data scientists,” he says.

The cloud poses its own challenges, though. There are cost and time considerations for moving data into and out of the cloud. The startup’s founder points out that data into Amazon is free; the cloud provider only charges for data out. On top of that, Yottamine doesn’t charge extra for their clients’ Amazon usage. The majority of users will use the Internet to upload their data to Amazon’s S3 storage system.

The models dictate the size of the data. Some are very compact, so getting them out of the cloud is not that expensive or time-consuming. But if someone had an enormous data set, they could use Amazon’s disk delivery system (aka sneakernet). However, the Yottamine reps do not anticipate many issues. “The data charges are small compared to the computational charges from an infrastructure cost standpoint,” observes Negris.

Regarding potential data concerns around security and privacy, the rep is quick to respond: “In our case, the data that’s actually being operated on – the data that’s being submitted to the SVM algorithm – is simply a binary matrix; it’s not a data set in any conventional sense. We stuff it into Hadoop, but it’s a binary matrix,” he says.

“Not only is it denatured, it goes beyond that. There’s a fundamental data transformation on the way to the cloud where what you’re handing up is a matrix of ones and zeros that has zero human-recognizable information or even machine-recognizable information, because the metadata is not necessary. Column heading information – ­ race, zip code or age, for example ­ – is not necessary to obtain the model. Somebody trying to get at the data would see a one in the first column, but wouldn’t know what that column was – there wouldn’t be a way to reverse engineer it.”

When it comes to Amazon EC2 instance types, Yottamine selects the virtual machine that best matches the problem size and the algorithm selected. The default for a linear algorithm that requires high CPU performance is the High-CPU Extra Large Instance. For nonlinear problems, which are more complex and require more memory even when the size of the problem is not very big, the company often employs Amazon’s High Memory Instance, specifically the High-Memory Quadruple Extra Large Instance.

Memory is key; in-memory computing achieves the biggest efficiencies. For the fastest possible configurations, Yottamine utilizes the EC2 High-I/O Instance which is backed by SSD for even more speed, but since this is still a newer instance type Amazon generally limits their use to two per user. The High-I/O Instance also includes 10 Gigabit Ethernet, so it’s a good fit for algorithms that rely on parallel programming models like MPI, where that high-speed interconnect counts most.

But speed isn’t always the ultimate priority for customers. Negris explains that it often depends on the vertical and the frequency at which the model is refreshed. The models used in credit card fraud detection do not refresh that often – perhaps on a quarterly basis. It’s a very rigid and well-established domain, and the information is well-defined. But the data coming from a click-stream could not be more opposite, so the Web advertising sector has high-turnover for model design. Thus they tend to be more speed-sensitive and less cost-sensitive.

As for likely competition, Negris divides potential challenges into two camps. He anticipates the most difficult market to crack will be current Oracle and SAS customers. These traditional software vendors offer an SVM algorithm as part of their portfolio. Yottamine, however, claims to offer a wider selection of predictive modeling software with six different SVM algorithms alone, which gives them the edge in some situations.

The other market penetration challenge comes from research algorithms that were initially built as part of an academic project, thesis or dissertation and were put into the open source stream, but Negris asserts that these aren’t very polished or industry-hardened since they were usually built for a single-purpose. They also require a certain amount of user expertise and experience.

Yottamine is hoping that its unique cloud-based approach and focus on performance, accuracy and easy of use will give its software the advantage over these conventional commercial and open source algorithms.

Subscribe to HPCwire's Weekly Update!

Be the most informed person in the room! Stay ahead of the tech trends with industy updates delivered to you every week!

TACC Helps ROSIE Bioscience Gateway Expand its Impact

April 26, 2017

Biomolecule structure prediction has long been challenging not least because the relevant software and workflows often require high end HPC systems that many bioscience researchers lack easy access to. Read more…

By John Russell

Messina Update: The U.S. Path to Exascale in 16 Slides

April 26, 2017

Paul Messina, director of the U.S. Exascale Computing Project, provided a wide-ranging review of ECP’s evolving plans last week at the HPC User Forum. Read more…

By John Russell

IBM, Nvidia, Stone Ridge Claim Gas & Oil Simulation Record

April 25, 2017

IBM, Nvidia, and Stone Ridge Technology today reported setting the performance record for a “billion cell” oil and gas reservoir simulation. Read more…

By John Russell

ASC17 Makes Splash at Wuxi Supercomputing Center

April 24, 2017

A record-breaking twenty student teams plus scores of company representatives, media professionals, staff and student volunteers transformed a formerly empty hall inside the Wuxi Supercomputing Center into a bustling hub of HPC activity, kicking off day one of 2017 Asia Student Supercomputer Challenge (ASC17). Read more…

By Tiffany Trader

HPE Extreme Performance Solutions

Remote Visualization Optimizing Life Sciences Operations and Care Delivery

As patients continually demand a better quality of care and increasingly complex workloads challenge healthcare organizations to innovate, investing in the right technologies is key to ensuring growth and success. Read more…

Groq This: New AI Chips to Give GPUs a Run for Deep Learning Money

April 24, 2017

CPUs and GPUs, move over. Thanks to recent revelations surrounding Google’s new Tensor Processing Unit (TPU), the computing world appears to be on the cusp of a new generation of chips designed specifically for deep learning workloads. Read more…

By Alex Woodie

Musk’s Latest Startup Eyes Brain-Computer Links

April 21, 2017

Elon Musk, the auto and space entrepreneur and severe critic of artificial intelligence, is forming a new venture that reportedly will seek to develop an interface between the human brain and computers. Read more…

By George Leopold

MIT Mathematician Spins Up 220,000-Core Google Compute Cluster

April 21, 2017

On Thursday, Google announced that MIT math professor and computational number theorist Andrew V. Sutherland had set a record for the largest Google Compute Engine (GCE) job. Sutherland ran the massive mathematics workload on 220,000 GCE cores using preemptible virtual machine instances. Read more…

By Tiffany Trader

NERSC Cori Shows the World How Many-Cores for the Masses Works

April 21, 2017

As its mission, the high performance computing center for the U.S. Department of Energy Office of Science, NERSC (the National Energy Research Supercomputer Center), supports a broad spectrum of forefront scientific research across diverse areas that includes climate, material science, chemistry, fusion energy, high-energy physics and many others. Read more…

By Rob Farber

Messina Update: The U.S. Path to Exascale in 16 Slides

April 26, 2017

Paul Messina, director of the U.S. Exascale Computing Project, provided a wide-ranging review of ECP’s evolving plans last week at the HPC User Forum. Read more…

By John Russell

ASC17 Makes Splash at Wuxi Supercomputing Center

April 24, 2017

A record-breaking twenty student teams plus scores of company representatives, media professionals, staff and student volunteers transformed a formerly empty hall inside the Wuxi Supercomputing Center into a bustling hub of HPC activity, kicking off day one of 2017 Asia Student Supercomputer Challenge (ASC17). Read more…

By Tiffany Trader

Groq This: New AI Chips to Give GPUs a Run for Deep Learning Money

April 24, 2017

CPUs and GPUs, move over. Thanks to recent revelations surrounding Google’s new Tensor Processing Unit (TPU), the computing world appears to be on the cusp of a new generation of chips designed specifically for deep learning workloads. Read more…

By Alex Woodie

NERSC Cori Shows the World How Many-Cores for the Masses Works

April 21, 2017

As its mission, the high performance computing center for the U.S. Department of Energy Office of Science, NERSC (the National Energy Research Supercomputer Center), supports a broad spectrum of forefront scientific research across diverse areas that includes climate, material science, chemistry, fusion energy, high-energy physics and many others. Read more…

By Rob Farber

Hyperion (IDC) Paints a Bullish Picture of HPC Future

April 20, 2017

Hyperion Research – formerly IDC’s HPC group – yesterday painted a fascinating and complicated portrait of the HPC community’s health and prospects at the HPC User Forum held in Albuquerque, NM. HPC sales are up and growing ($22 billion, all HPC segments, 2016). Read more…

By John Russell

Knights Landing Processor with Omni-Path Makes Cloud Debut

April 18, 2017

HPC cloud specialist Rescale is partnering with Intel and HPC resource provider R Systems to offer first-ever cloud access to Xeon Phi "Knights Landing" processors. The infrastructure is based on the 68-core Intel Knights Landing processor with integrated Omni-Path fabric (the 7250F Xeon Phi). Read more…

By Tiffany Trader

CERN openlab Explores New CPU/FPGA Processing Solutions

April 14, 2017

Through a CERN openlab project known as the ‘High-Throughput Computing Collaboration,’ researchers are investigating the use of various Intel technologies in data filtering and data acquisition systems. Read more…

By Linda Barney

DOE Supercomputer Achieves Record 45-Qubit Quantum Simulation

April 13, 2017

In order to simulate larger and larger quantum systems and usher in an age of “quantum supremacy,” researchers are stretching the limits of today’s most advanced supercomputers. Read more…

By Tiffany Trader

Google Pulls Back the Covers on Its First Machine Learning Chip

April 6, 2017

This week Google released a report detailing the design and performance characteristics of the Tensor Processing Unit (TPU), its custom ASIC for the inference phase of neural networks (NN). Read more…

By Tiffany Trader

Quantum Bits: D-Wave and VW; Google Quantum Lab; IBM Expands Access

March 21, 2017

For a technology that’s usually characterized as far off and in a distant galaxy, quantum computing has been steadily picking up steam. Read more…

By John Russell

Trump Budget Targets NIH, DOE, and EPA; No Mention of NSF

March 16, 2017

President Trump’s proposed U.S. fiscal 2018 budget issued today sharply cuts science spending while bolstering military spending as he promised during the campaign. Read more…

By John Russell

HPC Compiler Company PathScale Seeks Life Raft

March 23, 2017

HPCwire has learned that HPC compiler company PathScale has fallen on difficult times and is asking the community for help or actively seeking a buyer for its assets. Read more…

By Tiffany Trader

Nvidia Responds to Google TPU Benchmarking

April 10, 2017

Nvidia highlights strengths of its newest GPU silicon in response to Google's report on the performance and energy advantages of its custom tensor processor. Read more…

By Tiffany Trader

CPU-based Visualization Positions for Exascale Supercomputing

March 16, 2017

In this contributed perspective piece, Intel’s Jim Jeffers makes the case that CPU-based visualization is now widely adopted and as such is no longer a contrarian view, but is rather an exascale requirement. Read more…

By Jim Jeffers, Principal Engineer and Engineering Leader, Intel

For IBM/OpenPOWER: Success in 2017 = (Volume) Sales

January 11, 2017

To a large degree IBM and the OpenPOWER Foundation have done what they said they would – assembling a substantial and growing ecosystem and bringing Power-based products to market, all in about three years. Read more…

By John Russell

TSUBAME3.0 Points to Future HPE Pascal-NVLink-OPA Server

February 17, 2017

Since our initial coverage of the TSUBAME3.0 supercomputer yesterday, more details have come to light on this innovative project. Of particular interest is a new board design for NVLink-equipped Pascal P100 GPUs that will create another entrant to the space currently occupied by Nvidia's DGX-1 system, IBM's "Minsky" platform and the Supermicro SuperServer (1028GQ-TXR). Read more…

By Tiffany Trader

Leading Solution Providers

Tokyo Tech’s TSUBAME3.0 Will Be First HPE-SGI Super

February 16, 2017

In a press event Friday afternoon local time in Japan, Tokyo Institute of Technology (Tokyo Tech) announced its plans for the TSUBAME3.0 supercomputer, which will be Japan’s “fastest AI supercomputer,” Read more…

By Tiffany Trader

Is Liquid Cooling Ready to Go Mainstream?

February 13, 2017

Lost in the frenzy of SC16 was a substantial rise in the number of vendors showing server oriented liquid cooling technologies. Three decades ago liquid cooling was pretty much the exclusive realm of the Cray-2 and IBM mainframe class products. That’s changing. We are now seeing an emergence of x86 class server products with exotic plumbing technology ranging from Direct-to-Chip to servers and storage completely immersed in a dielectric fluid. Read more…

By Steve Campbell

IBM Wants to be “Red Hat” of Deep Learning

January 26, 2017

IBM today announced the addition of TensorFlow and Chainer deep learning frameworks to its PowerAI suite of deep learning tools, which already includes popular offerings such as Caffe, Theano, and Torch. Read more…

By John Russell

BioTeam’s Berman Charts 2017 HPC Trends in Life Sciences

January 4, 2017

Twenty years ago high performance computing was nearly absent from life sciences. Today it’s used throughout life sciences and biomedical research. Genomics and the data deluge from modern lab instruments are the main drivers, but so is the longer-term desire to perform predictive simulation in support of Precision Medicine (PM). There’s even a specialized life sciences supercomputer, ‘Anton’ from D.E. Shaw Research, and the Pittsburgh Supercomputing Center is standing up its second Anton 2 and actively soliciting project proposals. There’s a lot going on. Read more…

By John Russell

HPC Startup Advances Auto-Parallelization’s Promise

January 23, 2017

The shift from single core to multicore hardware has made finding parallelism in codes more important than ever, but that hasn’t made the task of parallel programming any easier. Read more…

By Tiffany Trader

Facebook Open Sources Caffe2; Nvidia, Intel Rush to Optimize

April 18, 2017

From its F8 developer conference in San Jose, Calif., today, Facebook announced Caffe2, a new open-source, cross-platform framework for deep learning. Caffe2 is the successor to Caffe, the deep learning framework developed by Berkeley AI Research and community contributors. Read more…

By Tiffany Trader

HPC Technique Propels Deep Learning at Scale

February 21, 2017

Researchers from Baidu’s Silicon Valley AI Lab (SVAIL) have adapted a well-known HPC communication technique to boost the speed and scale of their neural network training and now they are sharing their implementation with the larger deep learning community. Read more…

By Tiffany Trader

US Supercomputing Leaders Tackle the China Question

March 15, 2017

Joint DOE-NSA report responds to the increased global pressures impacting the competitiveness of U.S. supercomputing. Read more…

By Tiffany Trader

  • arrow
  • Click Here for More Headlines
  • arrow
Share This