Artificial intelligence (AI) is the buzz across a wide range of industries due to its seemingly endless possibilities. Machine learning is a subset of artificial intelligence which gives computers the capacity to learn tasks without being explicitly programmed to do so. Companies worldwide are focusing on specific ways they can integrate machine learning solutions into their infrastructure.
Customers from a variety of industries have shared their ideas with Dell EMC and Intel - from automobile manufacturers using optimized frameworks like Torch*, Caffe* and Theano* for image recognition in autonomous vehicles, to financial services using TensorFlow* for risk evaluation and fraud detection.
It’s exciting to listen to customers talk about how they plan to change industries and disrupt markets. Reality sets in when customers begin executing the proof of concept. Machine learning hype focuses on the end result. But, building the complete stack is very complex, and many customers are struggling to realize their vision.
Specifically, our customers face a number of challenges:
- Deploying, maintaining, and optimizing machine learning environments requires extensive human resources.
- Limited expertise and quickly-changing technology requires agility and the ability to tune and optimize without reducing stability.
- The market is fragmented, with each vendor in the stack strictly focused on their piece and little integration between hardware, software, frameworks, and libraries.
- Each ‘turning of the knob’ can lead to failed jobs, create incompatible software versions, or lack the ability to take advantage of accelerated compute.
Enterprise customer deployments for machine learning are still in the early phases. Many early customers are encountering significant challenges. That’s because machine learning is quite complicated to get right. The complexity comes from the quality of datasets being used to train, tuning to increase accuracy, framework optimizations, so on and so forth.
Apart from this, machine learning is an iterative process, and data scientists need to run through the steps multiple times to achieve the required accuracy. Additionally, customers have to consider data gravity and when to co-locate machine learning analysis to the data for performance. All this equals frustration and lack of customer adoption.
Dell EMC and Intel want to simplify this process and make it easier for customers to take advantage of machine learning. We believe that a machine learning solution should be easy to implement, manage, and maintain, without asking our customers to learn a wide variety of new skills.
To that end, Dell EMC and Intel have partnered to help our customers streamline their machine learning deployments. We simplify the stack from hardware to software implementation, with interoperability across the solution. We can help customers bridge the skills gaps in infrastructure, software, models, algorithms, and data science.
New Innovations Make Machine Learning Accessible
One example of the Dell EMC and Intel collaboration is the newly-announced, co-engineered Dell EMC Ready Bundle for Machine Learning featuring Intel-optimized BigDL. This solution provides an architecture optimized and tuned for image recognition and recommendation. The stack will include hardware, software, pre-trained models, and services including deployment, integration, and tuning based on use case.
The Dell EMC Ready Bundle for Machine Learning is built on a solid foundation anchored by PowerEdge R740XD/R640 and featuring Intel® Xeon® Scalable processors for high-performance machine learning training and inference workloads. It also integrates BigDL, an open source distributed deep learning library for Apache Spark* developed by Intel and allowing the ability to add deep learning to applications on the same Hadoop/Spark cluster where the data is stored. This makes deep learning more accessible for big data users and data scientists, who are usually not experts in deep learning.
We recognize customers will differ in skill set and expertise, so to help customers bridge this gap, Dell EMC will provide a flexible set of tools to simplify the building of a machine learning data pipeline. Partnering with Cloudera, we are able to use the Cloudera Data Science Workbench, which enables fast, easy, and secure self-service data science for the enterprise. Data scientists manage their own analytics pipelines, including built-in scheduling, monitoring, and email alerting.
We are also working with DataRobot, whose technology transforms model building by truly democratizing the process — perhaps the most important element of any enterprise machine learning platform. It also automates the entire modeling lifecycle, enabling users to quickly and easily build highly accurate predictive models.
Bridging the Skills Gap
Another differentiator is our expertise. Dell EMC subject matter experts in data science, data engineering, and analytics are available to help bridge the skills gap for customers. Dell EMC Consulting can provide fixed scope service for use cases tailored for fraud detection and image recognition. Our expertise spans from ingest to insights with full data pipeline architecture expertise, resulting in faster time to value.
Intrigued? Come check us out at SC17 and see how we can help you understand if machine learning makes sense for your use case. Please also visit the Dell High-Performance Computing page to learn more and follow the conversation along online with @DellEMC and @IntelHPC.