Edit: Some links no longer work.
Originally posted January 23, 2018 on AIXchange
I found this IBM developerWorks post about PowerAI on the IBM Linux on Power Twitter feed (@ibmpowerlinux).
This information is a pleasant surprise. Articulating why customers should care about PowerAI can be challenging. In many cases this workload is handled by departments or organizations that are different from the ones we typically work with:
PowerAI is an IBM Cognitive Systems offering for the rapidly growing and quickly evolving artificial intelligence (AI) category of deep learning. PowerAI brings a suite of capabilities from the open source community and combines them into a single enterprise distribution of software that incorporates complete lifecycle management from installation and configuration; data ingest and preparation; building, optimizing, and training the model; to inference; testing; and moving the model into production.
Busy as we are tending to AIX servers and workloads, topics like TensorFlow or Caffe seldom come up. We might skim articles about AI or deep learning, but we quickly move on. But this post connects the dots for us:
Deep learning is the fastest growing subcategory of machine learning and uses software neural networks to help develop patterns of analysis within the system to generate predictive capability: deep learning is a platform that is capable of effectively learning how to learn, and it is immensely powerful for helping clients get the most out of their data.
You may think this information applies only to some distant future, but I find it quite timely. Look at it this way: Things are very different in our data centers today compared to 20 years ago. We need to have an idea of what’s coming over the next 20 years: What will PowerAI give your organization?
- Helps to make deep learning easier and faster for organizations….
- Designed to provide an end-to-end deep learning platform for data scientist.
- Ready-to-use deep learning frameworks (TensorFlow, IBM Caffe, and BVLC Caffe).
- Distributed as easy-to-install binaries.
- Includes all dependencies and libraries.
- Easy updates: Code updates arrive from a repository….
- Designed for enterprise scale. PowerAI enables clients to distribute the training of a model across many servers, with the potential for greatly improving performance. What used to take weeks on a single server can potentially now be completed in just hours. This distributed capability is also transparent to the application logic previously written for a single server implementation. It is the best of both worlds: potential performance improvements without having to change the application code.
- Deep learning to unleash new analytic capabilities….
- Training neural network models. With PowerAI, data scientists have visual tools for understanding accuracy while the model is running; If accuracy is not high, the model can be stopped without wasting additional time.
IBM intends to deliver IBM PowerAI Vision, an application development tool for computer vision workloads. IBM PowerAI Vision is intended to automatically train deep learning models for different image and video input data sets.
Also check out these PowerAI videos: a shorter version, a longer version and an installation how-to.