POWER9 is 'Google Strong'
At the recent OpenPOWER Summit in Las Vegas, Google's Maire Mahony declares the #POWER9-based Zaius platform to be "Google strong"!
Which is not only really cool, it also shows the passion they have for the platform and how they believe it will help them go beyond Moores Law.
If you'd like to see the full talk, watch this video
Google believe in P9
AI Vision in Action
PowerAI Vision and IVA
IBM's head of Machine Learning/Deep Learning products, Sumit Gupta, announced that IBM PowerAI Vision and IVA would be combined to enable developers to start even faster!
Intelligent Video Analytics provides 'out of the box' video analytics capabilities which can now be enhanced via the AI Vision toolset to enable you to go even deeper for your specific use cases
If you'd like to see Sumit's blog check out this link
How to Accelerate the Use of AI in Organizations
Open any business publication or digital journal today, and you will read about the promise of AI, known as artificial or augmented intelligence, and how it will transform your business. The fact is, AI will not only transform your entire business, whether you are in health care, finance, retail or manufacturing, but it will also transform technology itself. more on ibm.com
Let’s discuss with Jean-Armand Broyelle, the technical Leader of the Cognitive Systems Lab, IBM Montpellier - France .
Can you tell us more on who you are and what have been your journey to AI/Cognitive?
I have 48 years, married, 2 kids, fond of sciences and nature sports, and I work as IT Engineer for IBM since 1998 when they acquired Sequent. After years as Power Systems Benchmark Expert, I had the opportunity to explore the world of accelerated workload through CAPI, the IBM technology for FPGA acceleration. Then I developped my skills in Deep Learning thanks to the IBM Intelligent Video Analytics (IVA) solution that we ported & tuned to Power Systems with GPUs and the PowerAI tool-box, end of 2015. I then joined the IBM European Cognitive Systems team led by I.Peacock mid-2017.
Can you tell us where you work & what is your exact job role?
Our mission at the IBM Montpellier Cognitive Systems Lab is to help IBM customers & partners to realize the full value of Power Systems. We are a team of 12 Technical Specialists that master Power & OpenPower systems running Linux. I focus with my team on AI Solutions such as PowerAI, PowerAI Vision, DSX and Deep Learning Impact from Conductor with Spark.
If I am an AI user, How could you and your team help me?
Simple, contact us! In a first call, we will discuss your technical and business context, your level in ML/DL and your objectives. We’ll then go to the next level through different kind of workshops where we could present and demo products, explore specific AI scenarios or allow you to experiment PowerAI through hands-on! And if you want more, like experimenting prototypes or testing, guess what ? We like to get our “hands dirty” and demonstrate the benefits of our platforms on real hardware and realistic workloads.
So you can also support AI PoC and Benchmarks?
Yes we have dedicated servers on which we can organize you hands-on fast-start skill transfer, enabling your team to leverage Power Systems &PowerAI and make them autonomous to install, optimize, profile and train models. It is clearly one of our best seller offerings: we did 2 in Q1 for UK & Hungarian partners and already have 3 in plan for Q2 with 2 Swiss & 1 Finish customers. PoCs and Benchmark are our speciality at Montpellier: we love execute them (and win ;-). It comes from our HPC team culture for application performance optimization since 2000: monitor/profile/tune workloads, adapt/modify source code, analyze/experiment technical applications is part of our DNA.
What is your best souvenir?
Last year during a PoC, I spent four weeks adapting Keras+Tensorflow code to enable PowerAIdistributed-deep-learning (DDL) for a Belgian customer which use CNNs models for electronic parts VIQ scenarios. You can find more via our client center web site: https://www.ibm.com/ibm/clientcenter/montpellier/index.shtml
You worked apparently on many projects? What are for you the 3 key lessons you want to share with the community on AI?
1) The AI Training platform will make the difference at some point in time in your AI journey. We have worked with several companies that have built their business model on building AI solutions and they were complaining now that the training duration is the limitation to their imagination and ability to deliver the relevant models to their customers within the right timing: they don’t have the time to explore many models options because the data acquisition and preparation is long, training last too long and the essential “hyperparameters optimization” phase is tedious. Power with its unique capabilities (Power9, DDL, LMS, NVlink, DLI...) can really solve these issues without compromise with the models accuracy.
2) AI will really succeed if it’s easily accessible to everyone. Nowardays, some DL technics are becoming main-streams (CNNs, R-CNNs, Auto encoders, ...) and lot of AI use-case don’t need the assistance of a Datascientist dedicated full-time to the project. It means that non-IT SMEs should be autonomous to develop some AI solutions through all the mandatory steps: manage data, develop and tune models, evaluate and deploy them. With Cognitive Systems, we have solutions for that like PowerAI Vision!
3) The “Data is the new oil” quote is true under different point of view. The mature companies that transform their data in meaningful models for new business applications are protecting their data and DL models like in-house IP. Some of them even monetize the access to their data. It means that the securization and life cycle of your data are really two important things to consider when building your AI infrastructure. Lot of organization which begins in AI underestimate it. “Data is the new oil” implies “Protect your data“. Again, we can help you to manage this new paradigm with our solutions.
If we want to team with you and your team, how do we proceed?
The best is to send me a mail directly at email@example.com copying the Cognitive Systems lab coordinator Alain Roy (firstname.lastname@example.org).
Want to be featured? Fill out the community spotlight form and tell us what you've been up to!