Artificial Intelligence

  1. Home
  2. >
  3. Solutions
  4. >
  5. Artificial Intelligence

Overview

ComnetCo offers HPE & NVIDIA AI-optimized clusters. Running NVIDIA NGC optimized software, these solutions provide an extraordinarily powerful AI supercomputing infrastructure. This solution offers the compute, storage, networking, and software elements to handle the most complex AI business problems.

Let us help you Explore, Experiment, and Expand as you grow your organizational AI practice.

Machine Learning

Having proven remarkably successful in applications such as image recognition and object detection, Machine Learning (ML) comprises a fundamental building block of Artificial Intelligence.  We are seeing a steady increase in the level of our customer’s sophistication and experimentation with ML for a widening range of use cases. From atomic-level particle picking for Cryo-EM and the Life Sciences to analyzing artists’ brush strokes in the Fine Arts, ML is augmenting and often improving the way we perform some tasks.  But what exactly is Machine Learning?

ML starts with a model of how things work, then uses algorithms that can learn. This learning is cycled back into the model to improve its representational value and the understanding it conveys. ML does not rely on detailed specific instructions on what to do and how to do things. Instead, its algorithms compare the results obtained by using various methods and making specific choices. Over time, by repeatedly choosing methods and options that produce the best results, the model evolves and improves through a continuous cycle. To achieve this, Machine Learning, sometimes referred to as Centralized Learning, requires two components:

Data Collection – Usually at the edge

Data Ingestion – At the core data center, where training of the model is performed, and then the model is moved back to the edge for inferencing to produce results

Deep Learning

Where Machine Learning uses algorithms and data sets to create and refine its own self-generated understandings and techniques, Deep Learning (DL) layers complex hierarchical models intended to reflect or represent the human thought process.

Deep Learning is a subset of machine learning that has demonstrated significantly superior performance compared with some traditional machine learning approaches. It combines multi-layer artificial neural networks and data- and compute-intensive training inspired by our latest understanding of human brain behavior. Deep learning has become so effective that it has even begun to surpass human abilities in many areas, such as image and speech recognition and natural language processing.

Together with NVIDIA & HPE, ComnetCo offers a leading portfolio of optimized AI and deep learning solutions. We enable deep learning through online and instructor-led workshops, reference architectures, and benchmarks on NVIDIA GPU-accelerated applications. Our solutions are differentiated by proven expertise, the largest deep learning ecosystem, and AI software frameworks.

Swarm Learning

Effiency

Swarm Learning eliminates the costs of raw data transfer by performing learning at or near the data sources. Our experiments show that the model parameters transferred can be more than 1000x smaller than the raw data. This tremendously lowers the data transfer costs and delays.

Privacy and Security Compliance

Swarm Learning helps businesses comply with privacy and security regulations by giving data owners greater control over access to and usage of their data through the smart contract of blockchain, as well as eliminating the need for raw data transfer. Strong compliance with security and privacy regulations boosts customer confidence, and in turn, brings a business more revenue.

Fault Tolerance

Compared with the centralized machine learning approach, Swarm Learning decentralizes both data storage and learning, thereby avoiding a single point of failure. The Swarm Learning algorithm is effective in handling biased and unbalanced data at the various sources, and the smart contract can robustly handle exceptions, such as the lost connection of a data source to its Swarm Learning peers.

Timely Insights

Swarm Learning also reduces the latency between the creation of data and the availability of actionable insight derived from that data. With Swarm Learning, model retraining can be initiated as soon as new data becomes available at any data source. The learning captured can be shared immediately with all the Swarm Learning peers—without waiting for the data to be transferred, consolidated, and then mined. A shorter path between data and insights delivers benefits such as faster and more accurate responses to the ever-changing market.

Getting Started with AI

ComnetCo understands that AI, data, and analytics initiatives run along a continuum driven by business needs and goals. Each organization has its unique path towards building a data foundation, developing advanced analytics solutions, and experimenting with AI for select use cases.

Take the first step on your AI journey with a one-day workshop for key data, business, and IT stakeholders. Depending on your needs and goals, our senior AI and data experts will help you:

  • Examine and explore possible use cases
  • Explore data strategy to support use cases
  • Identify AI and analytics functionalities to reach your objectives
  • Get infrastructure and architecture recommendations
  • Define a high-level roadmap

NVIDIA-Certified Solutions

Simplifies Deployment of Accelerated Computing at Scale

NVIDIA-Certified HPE Systems bring together HPE servers, NVIDIA GPUs and NVIDIA networking in optimized configurations that are validated for performance, manageability, security, and scalability and are backed by enterprise-grade support from NVIDIA and HPE. 

With NVIDIA-Certified Systems™, enterprises can confidently choose performance-optimized hardware solutions to power their accelerated computing workloads—in small configurations and at scale. 

The NVIDIA-Certified Systems program encompasses a wide range of enterprise GPUs, and the latest networking smart network interface cards (SmartNICs) from NVIDIA. The certification test suite exercises the performance and functionality on a set of software that represents a wide range of real-world applications. This includes deep learning training, AI inference, data science algorithms, intelligent video analytics (IVA), high-performance computing (HPC), CUDA®, and rendering. It also covers infrastructure performance acceleration, such as network and storage offload, security, and remote management. 

NVIDIA-Certified HPE Systems provide the foundation for a complete enterprise accelerated computing platform. These servers provide industry-leading performance for AI, data analytics, and visualization applications, and they can be managed with IT frameworks from industry leaders such as VMware and Red Hat.

HPE Cray AI Development Environment

Build Better AI Models, Faster

For engineers and data scientists, Machine Learning (ML) comprises a never-ending quest for new solutions that will enable them to better focus on innovation and accelerate their time to production. The HPE Cray AI Development Environment is all about delivering better ML.

As your organization seeks greater opportunities for growth, the HPE Cray AI Development Environment can help you find them. With this future-focused platform, you can successfully operate in the Exascale era of high-performance computing. to broaden your AI and machine learning scale to remove the complexity and cost associated with ML model development.

By removing the complexity and cost associated with ML model development, this comprehensive platform speeds time to value for model developers by:

  • Removing the need to write infrastructure code
  • Making it easier for IT administrators to set up, manage, secure, and share AI compute clusters

With the HPE Cray AI Development Environment, ML practitioners can:

  • Train models faster using state-of-the-art distributed training—without changing their model code
  • Automatically find high-quality models with advanced hyperparameter tuning from the creators of state-of-the-art tuning algorithms such as Hyperband
  • Get more from their GPUs with smart scheduling, as well as reduce cloud GPU costs by seamlessly using spot instances
  • Track and reproduce their work with experiment tracking that works out of the box, covering code versions, metrics, checkpoints, and hyperparameters

Using a comprehensive array of features integrated into an easy-to-use, high-performance ML environment, ML engineers can focus on building better models—rather than managing IT infrastructure.