Serdar Yegulalp

About the Author Serdar Yegulalp


ONNX makes machine learning models portable, shareable

Read more 0 Comments

ONNX makes machine learning models portable, shareable

Read more 0 Comments

Mesosphere taps Kubernetes for container orchestration

Read more 0 Comments

Microsoft linker tool shrinks .Net applications

A long-requested and long-unfulfilled feature for .Net has finally been delivered by Microsoft and the Mono team: A linker that allows .Net applications to be stripped down to include only the parts of libraries that are actually used by the program at runtime.

The IL Linker project works by analyzing a .Net application and determining which libraries are never called by the application in question. “It is effectively an application-specific dead code analysis,” says Microsoft in its GitHub announcement for the project.

A long-term mission for IL Linker is to make it into “the primary linker for the .Net ecosystem.”

To read this article in full or to leave a comment, please click here

Read more 0 Comments

3 projects lighting a fire under machine learning

Mention machine learning, and many common frameworks pop into mind, from “old” stalwarts like Scikit-learn to juggernauts like Google’s TensorFlow. But the field is large and diverse, and useful innovations are bubbling up across the landscape.

Recent releases from three open source projects continue the march toward making machine learning faster, more scalable, and easier to use. PyTorch and Apache MXNet bring GPU support to machine learning and deep learning in Python. 

To read this article in full or to leave a comment, please click here

Read more 0 Comments

Pivotal, VMware team up to deploy Kubernetes on vSphere

Pivotal and VMware have teamed up to deliver commercial-grade Kubernetes distributions on both VMware vSphere and Google Cloud Platform (GCP).

Pivotal Container Service (PKS), launching in Q4 2017, runs Kubernetes atop VMware’s infrastructure management tools—vSphere, vSAN, and NSX. It also taps a project from Cloud Foundry, Kubo, originally created by Pivotal and Google, to deploy and manage Kubernetes on VMware’s stack.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

Microsoft’s Project Brainwave accelerates deep learning in Azure

Earlier this year, Google unveiled its Tensor Processing Unit, custom hardware for speeding up prediction-making with machine learning models.

Now Microsoft is trying something similar, with its Project Brainwave hardware, which supports many major deep learning systems in wide use. Project Brainwave covers many of the same goals as Google’s TPU: Speed up how predictions are served from machine learning models (in Brainwave case, those hosted in Azure, using custom hardware deployed in Microsoft’s cloud at scale). 

To read this article in full or to leave a comment, please click here

Read more 0 Comments

13 frameworks for mastering machine learning

13 frameworks for mastering machine learning
13 frameworks for mastering machine learning

Image by W.Rebel via Wikimedia

Over the past year, machine learning has gone mainstream with a bang. The “sudden” arrival of machine learning isn’t fueled by cheap cloud environments and ever more powerful GPU hardware alone. It is also due to an explosion of open source frameworks designed to abstract away the hardest parts of machine learning and make its techniques available to a broad class of developers.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

Docker Enterprise now runs Windows and Linux in one cluster

With the newest Docker Enterprise Edition, you can now have Docker clusters composed of nodes running different operating systems.

Three of the key OSes supported by Docker—Windows, Linux, and IBM System Z—can run applications side by side in the same cluster, all orchestrated by a common mechanism.

Clustering apps across multiple OSes in Docker requires that you build per-OS images for each app. But those apps, when running on both Windows and Linux, can be linked to run in concert via Docker’s overlay networking.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

Docker Enterprise now runs Windows and Linux in one cluster

With the newest Docker Enterprise Edition, you can now have Docker clusters composed of nodes running different operating systems.

Three of the key OSes supported by Docker—Windows, Linux, and IBM System Z—can run applications side by side in the same cluster, all orchestrated by a common mechanism.

Clustering apps across multiple OSes in Docker requires that you build per-OS images for each app. But those apps, when running on both Windows and Linux, can be linked to run in concert via Docker’s overlay networking.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

Docker Enterprise now runs Windows and Linux in one cluster

With the newest Docker Enterprise Edition, you can now have Docker clusters composed of nodes running different operating systems.

Three of the key OSes supported by Docker — Windows, Linux, and IBM System Z — can run applications side by side in the same cluster, all orchestrated by a common mechanism.

Clustering apps across multiple OSes in Docker requires that you build per-OS images for each app. But those apps, when running on both Windows and Linux, can be linked to run in concert via Docker’s overlay networking.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

Amazon joins Kubernetes-focused CNCF industry group

The Cloud Native Computing Foundation, created to promote and develop technologies like Kubernetes and core components of the container ecosystem spawned by Docker, welcomed Amazon Web Services into its fold this week.

Amazon comes on board as a top-level (“platinum”) member. According to Amazon’s Adrian Cockcroft, now a member of the CNCF’s governing board, containers are the big reason Amazon’s getting involved—at least, initially.

Amazon already has a major investment in container tech. Its ECS service provides managed containers that run via machine images deployed on clusters of EC2 instances. Its older Elastic Beanstalk service can deploy and manage Docker containers, although they’re scaled and managed via Amazon’s own internal stack, not the CNCF’s Kubernetes. And users can always manually deploy Docker Enterprise Edition, a container-centric Linux such as CoreOS, or a Kubernetes cluster on EC2.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

IBM speeds deep learning by using multiple servers

For everyone frustrated by how long it takes to train deep learning models, IBM has some good news: It has unveiled a way to automatically split deep-learning training jobs across multiple physical servers — not just individual GPUs, but whole systems with their own separate sets of GPUs.

Now the bad news: It’s available only in IBM’s PowerAI 4.0 software package, which runs exclusively on IBM’s own OpenPower hardware systems.

Distributed Deep Learning (DDL) doesn’t require developers to learn an entirely new deep learning framework. It repackages several common frameworks for machine learning: TensorFlow, Torch, Caffe, Chainer, and Theano. Deep learning projecs that use those frameworks can then run in parallel across multiple hardware nodes.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

IBM speeds deep learning by using multiple servers

For everyone frustrated by how long it takes to train deep learning models, IBM has some good news: It has unveiled a way to automatically split deep-learning training jobs across multiple physical servers — not just individual GPUs, but whole systems with their own separate sets of GPUs.

Now the bad news: It’s available only in IBM’s PowerAI 4.0 software package, which runs exclusively on IBM’s own OpenPower hardware systems.

Distributed Deep Learning (DDL) doesn’t require developers to learn an entirely new deep learning framework. It repackages several common frameworks for machine learning: TensorFlow, Torch, Caffe, Chainer, and Theano. Deep learning projecs that use those frameworks can then run in parallel across multiple hardware nodes.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

IBM speeds deep learning by using multiple servers

For everyone frustrated by how long it takes to train deep learning models, IBM has some good news: It has unveiled a way to automatically split deep-learning training jobs across multiple physical servers — not just individual GPUs, but whole systems with their own separate sets of GPUs.

Now the bad news: It’s available only in IBM’s PowerAI 4.0 software package, which runs exclusively on IBM’s own OpenPower hardware systems.

Distributed Deep Learning (DDL) doesn’t require developers to learn an entirely new deep learning framework. It repackages several common frameworks for machine learning: TensorFlow, Torch, Caffe, Chainer, and Theano. Deep learning projecs that use those frameworks can then run in parallel across multiple hardware nodes.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

IBM speeds deep learning by using multiple servers

For everyone frustrated by how long it takes to train deep learning models, IBM has some good news: It has unveiled a way to automatically split deep-learning training jobs across multiple physical servers — not just individual GPUs, but whole systems with their own separate sets of GPUs.

Now the bad news: It’s available only in IBM’s PowerAI 4.0 software package, which runs exclusively on IBM’s own OpenPower hardware systems.

Distributed Deep Learning (DDL) doesn’t require developers to learn an entirely new deep learning framework. It repackages several common frameworks for machine learning: TensorFlow, Torch, Caffe, Chainer, and Theano. Deep learning projecs that use those frameworks can then run in parallel across multiple hardware nodes.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

Apache Spark 2.2 gets streaming, R language boosts

With version 2.2 of Apache Spark, a long-awaited feature for the multipurpose in-memory data processing framework is now available for production use.

Structured Streaming, as that feature is called, allows Spark to process streams of data in ways that are native to Spark’s batch-based data-handling metaphors. It’s part of Spark’s long-term push to become, if not all things to all people in data science, then at least the best thing for most of them.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

Apache Spark 2.2 gets streaming, R language boosts

With version 2.2 of Apache Spark, a long-awaited feature for the multipurpose in-memory data processing framework is now available for production use.

Structured Streaming, as that feature is called, allows Spark to process streams of data in ways that are native to Spark’s batch-based data-handling metaphors. It’s part of Spark’s long-term push to become, if not all things to all people in data science, then at least the best thing for most of them.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

What’s new in Kubernetes 1.7

In Kubernetes version 1.7, the container orchestration and management system is gaining slew of new security, stateful application, and extensibility features

Kubernetes 1.6 was mainly about stabilizing and bringing to term long-planned changes, such as using version 3 of the ETCD distributed key-value store. But many of Kubernetes 1.7’s new features are only in the alpha stage, more signals of how Kubernetes is trying to be more useful in a broader range of scenarios. Other new capabilities bring in features previously relegated to other parts of the container ecosystem.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

Nvidia’s new TensorRT speeds machine learning predictions

Nvidia has released a new version of TensorRT, a runtime system for serving inferences using deep learning models through Nvidia’s own GPUs.

Inferences, or predictions made from a trained model, can be served from either CPUs or GPUs. Serving inferences from GPUs is part of Nvidia’s strategy to get greater adoption of its processors.

Nvidia claims the GPU-based TensorRT is better across the board for inferencing than CPU-only approaches. One of Nvidia’s proffered benchmarks, the AlexNet image classification test under the Caffe framework, claims TensorRT to be 42 times faster than a CPU-only version of the same test — 16,041 images per second vs. 374—when run on Nvidia’s Tesla P40 processor. (Always take industry benchmarks with a grain of salt.)

To read this article in full or to leave a comment, please click here

Read more 0 Comments

Nvidia’s new TensorRT speeds machine learning predictions

Nvidia has released a new version of TensorRT, a runtime system for serving inferences using deep learning models through Nvidia’s own GPUs.

Inferences, or predictions made from a trained model, can be served from either CPUs or GPUs. Serving inferences from GPUs is part of Nvidia’s strategy to get greater adoption of its processors, countering what AMD is doing to break Nvidia’s stranglehold on the machine learning GPU market.

Nvidia claims the GPU-based TensorRT is better across the board for inferencing than CPU-only approaches. One of Nvidia’s proffered benchmarks, the AlexNet image classification test under the Caffe framework, claims TensorRT to be 42 times faster than a CPU-only version of the same test — 16,041 images per second vs. 374—when run on Nvidia’s Tesla P40 processor. (Always take industry benchmarks with a grain of salt.)

To read this article in full or to leave a comment, please click here

Read more 0 Comments

Nvidia’s new TensorRT speeds machine learning predictions

Nvidia has released a new version of TensorRT, a runtime system for serving inferences using deep learning models through Nvidia’s own GPUs.

Inferences, or predictions made from a trained model, can be served from either CPUs or GPUs. Serving inferences from GPUs is part of Nvidia’s strategy to get greater adoption of its processors, countering what AMD is doing to break Nvidia’s stranglehold on the machine learning GPU market.

Nvidia claims the GPU-based TensorRT is better across the board for inferencing than CPU-only approaches. One of Nvidia’s proffered benchmarks, the AlexNet image classification test under the Caffe framework, claims TensorRT to be 42 times faster than a CPU-only version of the same test — 16,041 images per second vs. 374—when run on Nvidia’s Tesla P40 processor. (Always take industry benchmarks with a grain of salt.)

To read this article in full or to leave a comment, please click here

Read more 0 Comments

Google open-sources TensorFlow training tools

Over the past year, Google’s TensorFlow has asserted itself as a popular open source toolkit for deep learning. But training a TensorFlow model can be cumbersome and slow—especially when the mission is to take a dataset used by someone else and try to refine the training process it uses. The sheer number of moving parts and variations in any model-training process is enough to make even deep-learning experts take a deep breath.

This week, Google open-sourced a project intended to cut down on the amount of work in configuring a deep learning model for training. Tensor2Tensor, or T2T for short, is a Python-powered workflow organization library for TensorFlow training jobs. It lets developers specify the key elements used in a TensorFlow model and define the relationships among them.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

How Heptio plans to automate away Kubernetes setup pains

Heptio, the commercial Kubernetes outfit founded by two creators of that microservices management framework, has unveiled its first public project for making Kubernetes easier to deploy in the enterprise.

Kubernetes simplifies how apps run as microservices, but setting up Kubernetes itself is no picnic. Heptio’s project automates some of the fiddlier parts of the setup process via a custom, domain-specific language.

Heptio’s project, dubbed Ksonnet, is an open source tool set for assembling the configuration needed to deploy Kubernetes. The most common setup difficulties in Kubernetes involve creating the configuration files, what Heptio calls the “wall of YAML” problem.

To read this article in full or to leave a comment, please click here

Read more 0 Comments

Atom editor now has native GitHub integration

GitHub’s Atom, the Node.js- and HTML5-powered code editor, has traditionally integrated with Git repositories—including GitHub itself—only by way of third-party components.

All that changed this week with the unveiling of a new core package for Atom, called appropriately enough GitHub for Atom, along with new release and beta editions of Atom itself.

GitHub users, dock here

The new edition of Atom, version 1.17, introduces a new UI component called “docks,” which is a way to provide side- or bottom-dockable tool panels in the editor. IDEs like Visual Studio and Eclipse have had dock-like components for some time, but now Atom is adding such a component as a core element.

To read this article in full or to leave a comment, please click here

Read more 0 Comments