�ݺ�ߣshows by User: NickPentreath

�ݺ�ߣshows by User: NickPentreath / http://www.slideshare.net/images/logo.gif �ݺ�ߣshows by User: NickPentreath / Wed, 18 Nov 2020 14:52:14 GMT �ݺ�ߣShare feed for �ݺ�ߣshows by User: NickPentreath Notebook-based AI Pipelines with Elyra and Kubeflow /slideshow/notebookbased-ai-pipelines-with-elyra-and-kubeflow/239318441 dataaisummiteu-2020-np-final-201118145214
A typical machine learning pipeline begins as a series of preprocessing steps followed by experimentation, optimization and model-tuning, and, finally deployment. Jupyter notebooks have become a hugely popular tool for data scientists and other machine learning practitioners to explore and experiment as part of this workflow, due to the flexibility and interactivity they provide. However, with notebooks it is often a challenge to move from the experimentation phase to creating a robust, modular and production-grade end-to-end AI pipeline. Elyra is a set of open-source, AI centric extensions to JupyterLab. Elyra provides a visual editor for building notebook-based pipelines that simplifies the conversion of multiple notebooks into batch jobs or workflows. These workflows can be executed both locally (during the experimentation phase) and on Kubernetes via Kubeflow Pipelines for production deployment. In this way, Elyra combines the flexibility and ease-of-use of notebooks and JupyterLab, with the production-grade qualities of Kubeflow (and in future potentially other Kubernetes-based orchestration platforms). In this talk I introduce Elyra and its capabilities, then give a deep dive of Elyra's pipeline editor and the underlying pipeline execution mechanics, showing a demo of using Elyra to construct an end-to-end analytics and machine learning pipeline. I will also explore how to integrate and scale out model-tuning as well as deployment via Kubeflow Serving.]]>
A typical machine learning pipeline begins as a series of preprocessing steps followed by experimentation, optimization and model-tuning, and, finally deployment. Jupyter notebooks have become a hugely popular tool for data scientists and other machine learning practitioners to explore and experiment as part of this workflow, due to the flexibility and interactivity they provide. However, with notebooks it is often a challenge to move from the experimentation phase to creating a robust, modular and production-grade end-to-end AI pipeline. Elyra is a set of open-source, AI centric extensions to JupyterLab. Elyra provides a visual editor for building notebook-based pipelines that simplifies the conversion of multiple notebooks into batch jobs or workflows. These workflows can be executed both locally (during the experimentation phase) and on Kubernetes via Kubeflow Pipelines for production deployment. In this way, Elyra combines the flexibility and ease-of-use of notebooks and JupyterLab, with the production-grade qualities of Kubeflow (and in future potentially other Kubernetes-based orchestration platforms). In this talk I introduce Elyra and its capabilities, then give a deep dive of Elyra's pipeline editor and the underlying pipeline execution mechanics, showing a demo of using Elyra to construct an end-to-end analytics and machine learning pipeline. I will also explore how to integrate and scale out model-tuning as well as deployment via Kubeflow Serving.]]> Wed, 18 Nov 2020 14:52:14 GMT /slideshow/notebookbased-ai-pipelines-with-elyra-and-kubeflow/239318441 NickPentreath@slideshare.net(NickPentreath) Notebook-based AI Pipelines with Elyra and Kubeflow NickPentreath A typical machine learning pipeline begins as a series of preprocessing steps followed by experimentation, optimization and model-tuning, and, finally deployment. Jupyter notebooks have become a hugely popular tool for data scientists and other machine learning practitioners to explore and experiment as part of this workflow, due to the flexibility and interactivity they provide. However, with notebooks it is often a challenge to move from the experimentation phase to creating a robust, modular and production-grade end-to-end AI pipeline. Elyra is a set of open-source, AI centric extensions to JupyterLab. Elyra provides a visual editor for building notebook-based pipelines that simplifies the conversion of multiple notebooks into batch jobs or workflows. These workflows can be executed both locally (during the experimentation phase) and on Kubernetes via Kubeflow Pipelines for production deployment. In this way, Elyra combines the flexibility and ease-of-use of notebooks and JupyterLab, with the production-grade qualities of Kubeflow (and in future potentially other Kubernetes-based orchestration platforms). In this talk I introduce Elyra and its capabilities, then give a deep dive of Elyra's pipeline editor and the underlying pipeline execution mechanics, showing a demo of using Elyra to construct an end-to-end analytics and machine learning pipeline. I will also explore how to integrate and scale out model-tuning as well as deployment via Kubeflow Serving. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/dataaisummiteu-2020-np-final-201118145214-thumbnail.jpg?width=120&height=120&fit=bounds" /> A typical machine learning pipeline begins as a series of preprocessing steps followed by experimentation, optimization and model-tuning, and, finally deployment. Jupyter notebooks have become a hugely popular tool for data scientists and other machine learning practitioners to explore and experiment as part of this workflow, due to the flexibility and interactivity they provide. However, with notebooks it is often a challenge to move from the experimentation phase to creating a robust, modular and production-grade end-to-end AI pipeline. Elyra is a set of open-source, AI centric extensions to JupyterLab. Elyra provides a visual editor for building notebook-based pipelines that simplifies the conversion of multiple notebooks into batch jobs or workflows. These workflows can be executed both locally (during the experimentation phase) and on Kubernetes via Kubeflow Pipelines for production deployment. In this way, Elyra combines the flexibility and ease-of-use of notebooks and JupyterLab, with the production-grade qualities of Kubeflow (and in future potentially other Kubernetes-based orchestration platforms). In this talk I introduce Elyra and its capabilities, then give a deep dive of Elyra's pipeline editor and the underlying pipeline execution mechanics, showing a demo of using Elyra to construct an end-to-end analytics and machine learning pipeline. I will also explore how to integrate and scale out model-tuning as well as deployment via Kubeflow Serving.

Notebook-based AI Pipelines with Elyra and Kubeflow from Nick Pentreath

]]> 66 0 https://cdn.slidesharecdn.com/ss_thumbnails/dataaisummiteu-2020-np-final-201118145214-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Scaling up deep learning by scaling down /slideshow/scaling-up-deep-learning-by-scaling-down/236164597 sparkaisummit-2020-np-200624190925
Spark Summit 2020 Talk In the last few years, deep learning has achieved dramatic success in a wide range of domains, including computer vision, artificial intelligence, speech recognition, natural language processing and reinforcement learning. However, good performance comes at a significant computational cost. This makes scaling training expensive, but an even more pertinent issue is inference, in particular for real-time applications (where runtime latency is critical) and edge devices (where computational and storage resources may be limited). This talk will explore common techniques and emerging advances for dealing with these challenges, including best practices for batching; quantization and other methods for trading off computational cost at training vs inference performance; architecture optimization and graph manipulation approaches.]]>
Spark Summit 2020 Talk In the last few years, deep learning has achieved dramatic success in a wide range of domains, including computer vision, artificial intelligence, speech recognition, natural language processing and reinforcement learning. However, good performance comes at a significant computational cost. This makes scaling training expensive, but an even more pertinent issue is inference, in particular for real-time applications (where runtime latency is critical) and edge devices (where computational and storage resources may be limited). This talk will explore common techniques and emerging advances for dealing with these challenges, including best practices for batching; quantization and other methods for trading off computational cost at training vs inference performance; architecture optimization and graph manipulation approaches.]]> Wed, 24 Jun 2020 19:09:24 GMT /slideshow/scaling-up-deep-learning-by-scaling-down/236164597 NickPentreath@slideshare.net(NickPentreath) Scaling up deep learning by scaling down NickPentreath Spark Summit 2020 Talk In the last few years, deep learning has achieved dramatic success in a wide range of domains, including computer vision, artificial intelligence, speech recognition, natural language processing and reinforcement learning. However, good performance comes at a significant computational cost. This makes scaling training expensive, but an even more pertinent issue is inference, in particular for real-time applications (where runtime latency is critical) and edge devices (where computational and storage resources may be limited). This talk will explore common techniques and emerging advances for dealing with these challenges, including best practices for batching; quantization and other methods for trading off computational cost at training vs inference performance; architecture optimization and graph manipulation approaches. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/sparkaisummit-2020-np-200624190925-thumbnail.jpg?width=120&height=120&fit=bounds" /> Spark Summit 2020 Talk In the last few years, deep learning has achieved dramatic success in a wide range of domains, including computer vision, artificial intelligence, speech recognition, natural language processing and reinforcement learning. However, good performance comes at a significant computational cost. This makes scaling training expensive, but an even more pertinent issue is inference, in particular for real-time applications (where runtime latency is critical) and edge devices (where computational and storage resources may be limited). This talk will explore common techniques and emerging advances for dealing with these challenges, including best practices for batching; quantization and other methods for trading off computational cost at training vs inference performance; architecture optimization and graph manipulation approaches.

Scaling up deep learning by scaling down from Nick Pentreath

]]> 186 0 https://cdn.slidesharecdn.com/ss_thumbnails/sparkaisummit-2020-np-200624190925-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 End-to-End Deep Learning Deployment with ONNX /slideshow/endtoend-deep-learning-deployment-with-onnx/234823747 onnxpipelines-deployai-mar2020-np-200601125812
The Open Neural Network Exchange (ONNX) standard has emerged for representing deep learning models in a standardized format. In this talk, I will discuss: 1. ONNX for exporting deep learning computation graphs, the ONNX-ML component of the specification for exporting both traditional ML models, common feature extraction, data transformation and post-processing steps. 2. How to use ONNX and the growing ecosystem of exporter libraries for common frameworks (including TensorFlow, PyTorch, Keras, scikit-learn and Apache SparkML) to deploy complete deep learning pipelines. 3. Best practices for working with and combining these disparate exporter toolkits, as well as highlight the gaps, issues, and missing pieces to be taken into account and still to be addressed.]]>
The Open Neural Network Exchange (ONNX) standard has emerged for representing deep learning models in a standardized format. In this talk, I will discuss: 1. ONNX for exporting deep learning computation graphs, the ONNX-ML component of the specification for exporting both traditional ML models, common feature extraction, data transformation and post-processing steps. 2. How to use ONNX and the growing ecosystem of exporter libraries for common frameworks (including TensorFlow, PyTorch, Keras, scikit-learn and Apache SparkML) to deploy complete deep learning pipelines. 3. Best practices for working with and combining these disparate exporter toolkits, as well as highlight the gaps, issues, and missing pieces to be taken into account and still to be addressed.]]> Mon, 01 Jun 2020 12:58:12 GMT /slideshow/endtoend-deep-learning-deployment-with-onnx/234823747 NickPentreath@slideshare.net(NickPentreath) End-to-End Deep Learning Deployment with ONNX NickPentreath The Open Neural Network Exchange (ONNX) standard has emerged for representing deep learning models in a standardized format. In this talk, I will discuss: 1. ONNX for exporting deep learning computation graphs, the ONNX-ML component of the specification for exporting both traditional ML models, common feature extraction, data transformation and post-processing steps. 2. How to use ONNX and the growing ecosystem of exporter libraries for common frameworks (including TensorFlow, PyTorch, Keras, scikit-learn and Apache SparkML) to deploy complete deep learning pipelines. 3. Best practices for working with and combining these disparate exporter toolkits, as well as highlight the gaps, issues, and missing pieces to be taken into account and still to be addressed. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/onnxpipelines-deployai-mar2020-np-200601125812-thumbnail.jpg?width=120&height=120&fit=bounds" /> The Open Neural Network Exchange (ONNX) standard has emerged for representing deep learning models in a standardized format. In this talk, I will discuss: 1. ONNX for exporting deep learning computation graphs, the ONNX-ML component of the specification for exporting both traditional ML models, common feature extraction, data transformation and post-processing steps. 2. How to use ONNX and the growing ecosystem of exporter libraries for common frameworks (including TensorFlow, PyTorch, Keras, scikit-learn and Apache SparkML) to deploy complete deep learning pipelines. 3. Best practices for working with and combining these disparate exporter toolkits, as well as highlight the gaps, issues, and missing pieces to be taken into account and still to be addressed.

End-to-End Deep Learning Deployment with ONNX from Nick Pentreath

]]> 259 0 https://cdn.slidesharecdn.com/ss_thumbnails/onnxpipelines-deployai-mar2020-np-200601125812-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Open, Secure & Transparent AI Pipelines /slideshow/open-secure-transparent-ai-pipelines/139616214 final-mlpipelines-anacondacon-apr19-np-190404185945
AI algorithms offer great promise in criminal justice, credit scoring, hiring and other domains. However, algorithmic fairness is a legitimate concern. Possible bias and adversarial contamination can come from training data, inappropriate data handling/model selection or incorrect algorithm design. This talk discusses how to build an open, transparent, secure and fair pipeline that fully integrates into the AI lifecycle — leveraging open-source projects such as AI Fairness 360 (AIF360), Adversarial Robustness Toolbox (ART), the Fabric for Deep Learning (FfDL) and the Model Asset eXchange (MAX).]]>
AI algorithms offer great promise in criminal justice, credit scoring, hiring and other domains. However, algorithmic fairness is a legitimate concern. Possible bias and adversarial contamination can come from training data, inappropriate data handling/model selection or incorrect algorithm design. This talk discusses how to build an open, transparent, secure and fair pipeline that fully integrates into the AI lifecycle — leveraging open-source projects such as AI Fairness 360 (AIF360), Adversarial Robustness Toolbox (ART), the Fabric for Deep Learning (FfDL) and the Model Asset eXchange (MAX).]]> Thu, 04 Apr 2019 18:59:45 GMT /slideshow/open-secure-transparent-ai-pipelines/139616214 NickPentreath@slideshare.net(NickPentreath) Open, Secure & Transparent AI Pipelines NickPentreath AI algorithms offer great promise in criminal justice, credit scoring, hiring and other domains. However, algorithmic fairness is a legitimate concern. Possible bias and adversarial contamination can come from training data, inappropriate data handling/model selection or incorrect algorithm design. This talk discusses how to build an open, transparent, secure and fair pipeline that fully integrates into the AI lifecycle — leveraging open-source projects such as AI Fairness 360 (AIF360), Adversarial Robustness Toolbox (ART), the Fabric for Deep Learning (FfDL) and the Model Asset eXchange (MAX). <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/final-mlpipelines-anacondacon-apr19-np-190404185945-thumbnail.jpg?width=120&height=120&fit=bounds" /> AI algorithms offer great promise in criminal justice, credit scoring, hiring and other domains. However, algorithmic fairness is a legitimate concern. Possible bias and adversarial contamination can come from training data, inappropriate data handling/model selection or incorrect algorithm design. This talk discusses how to build an open, transparent, secure and fair pipeline that fully integrates into the AI lifecycle — leveraging open-source projects such as AI Fairness 360 (AIF360), Adversarial Robustness Toolbox (ART), the Fabric for Deep Learning (FfDL) and the Model Asset eXchange (MAX).

Open, Secure & Transparent AI Pipelines from Nick Pentreath

]]> 316 4 https://cdn.slidesharecdn.com/ss_thumbnails/final-mlpipelines-anacondacon-apr19-np-190404185945-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 AI and Spark - IBM Community AI Day /slideshow/ai-and-spark-ibm-community-ai-day/119124421 ibm-aiday-sparkandai-oct18-np-181011145906
Apache Spark’s machine learning library provides a simple, elegant, yet powerful framework for creating scalable machine learning pipelines. It provides out of the box components for feature extraction and transformation, as well as various machine learning algorithms. However, in recent years specialized systems (such as TensorFlow, Caffe, PyTorch and Apache MXNet) have been dominant in the domain of AI and deep learning, as they allow greater performance and flexibility for training complex models. While there are a few deep learning frameworks that are Spark specific, in most cases these frameworks are separate from Spark and the ease of integration and feature set exposed varies considerably. This session will explore the role of Spark within the AI landscape, the current state of deep learning on top of Spark and the most recent developments in the Spark project to better integrate Spark with the deep learning ecosystem.]]>
Apache Spark’s machine learning library provides a simple, elegant, yet powerful framework for creating scalable machine learning pipelines. It provides out of the box components for feature extraction and transformation, as well as various machine learning algorithms. However, in recent years specialized systems (such as TensorFlow, Caffe, PyTorch and Apache MXNet) have been dominant in the domain of AI and deep learning, as they allow greater performance and flexibility for training complex models. While there are a few deep learning frameworks that are Spark specific, in most cases these frameworks are separate from Spark and the ease of integration and feature set exposed varies considerably. This session will explore the role of Spark within the AI landscape, the current state of deep learning on top of Spark and the most recent developments in the Spark project to better integrate Spark with the deep learning ecosystem.]]> Thu, 11 Oct 2018 14:59:06 GMT /slideshow/ai-and-spark-ibm-community-ai-day/119124421 NickPentreath@slideshare.net(NickPentreath) AI and Spark - IBM Community AI Day NickPentreath Apache Spark’s machine learning library provides a simple, elegant, yet powerful framework for creating scalable machine learning pipelines. It provides out of the box components for feature extraction and transformation, as well as various machine learning algorithms. However, in recent years specialized systems (such as TensorFlow, Caffe, PyTorch and Apache MXNet) have been dominant in the domain of AI and deep learning, as they allow greater performance and flexibility for training complex models. While there are a few deep learning frameworks that are Spark specific, in most cases these frameworks are separate from Spark and the ease of integration and feature set exposed varies considerably. This session will explore the role of Spark within the AI landscape, the current state of deep learning on top of Spark and the most recent developments in the Spark project to better integrate Spark with the deep learning ecosystem. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/ibm-aiday-sparkandai-oct18-np-181011145906-thumbnail.jpg?width=120&height=120&fit=bounds" /> Apache Spark’s machine learning library provides a simple, elegant, yet powerful framework for creating scalable machine learning pipelines. It provides out of the box components for feature extraction and transformation, as well as various machine learning algorithms. However, in recent years specialized systems (such as TensorFlow, Caffe, PyTorch and Apache MXNet) have been dominant in the domain of AI and deep learning, as they allow greater performance and flexibility for training complex models. While there are a few deep learning frameworks that are Spark specific, in most cases these frameworks are separate from Spark and the ease of integration and feature set exposed varies considerably. This session will explore the role of Spark within the AI landscape, the current state of deep learning on top of Spark and the most recent developments in the Spark project to better integrate Spark with the deep learning ecosystem.

AI and Spark - IBM Community AI Day from Nick Pentreath

]]> 120 2 https://cdn.slidesharecdn.com/ss_thumbnails/ibm-aiday-sparkandai-oct18-np-181011145906-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 IBM Developer Model Asset eXchange /slideshow/ibm-developer-model-asset-exchange/119113735 max-aiconf-oct18-np-181011124309
The common perception of applying deep learning is that you take an open source or research model, train it on raw data, and deploy the result as a fully self-contained artefact. The reality is far more complex. For the training phase, users face an array of challenges including handling varied deep learning frameworks, hardware requirements and configurations, not to mention code quality, consistency, and packaging. For the deployment phase, they face another set of challenges, ranging from custom requirements for data pre- and postprocessing, inconsistencies across frameworks, and lack of standardization in serving APIs. The goal of the IBM Developer Model Asset eXchange (MAX) is to remove these barriers to entry for developers to obtain, train, and deploy open source deep learning models for their business applications. In building the exchange, we encountered all these challenges and more. For the training phase, we leverage the Fabric for Deep Learning (FfDL), an open source project providing framework-independent training of deep learning models on Kubernetes. For the deployment phase, MAX provides standardized container-based, fully self-contained model artifacts encompassing the end-to-end deep learning predictive pipeline.]]>
The common perception of applying deep learning is that you take an open source or research model, train it on raw data, and deploy the result as a fully self-contained artefact. The reality is far more complex. For the training phase, users face an array of challenges including handling varied deep learning frameworks, hardware requirements and configurations, not to mention code quality, consistency, and packaging. For the deployment phase, they face another set of challenges, ranging from custom requirements for data pre- and postprocessing, inconsistencies across frameworks, and lack of standardization in serving APIs. The goal of the IBM Developer Model Asset eXchange (MAX) is to remove these barriers to entry for developers to obtain, train, and deploy open source deep learning models for their business applications. In building the exchange, we encountered all these challenges and more. For the training phase, we leverage the Fabric for Deep Learning (FfDL), an open source project providing framework-independent training of deep learning models on Kubernetes. For the deployment phase, MAX provides standardized container-based, fully self-contained model artifacts encompassing the end-to-end deep learning predictive pipeline.]]> Thu, 11 Oct 2018 12:43:09 GMT /slideshow/ibm-developer-model-asset-exchange/119113735 NickPentreath@slideshare.net(NickPentreath) IBM Developer Model Asset eXchange NickPentreath The common perception of applying deep learning is that you take an open source or research model, train it on raw data, and deploy the result as a fully self-contained artefact. The reality is far more complex. For the training phase, users face an array of challenges including handling varied deep learning frameworks, hardware requirements and configurations, not to mention code quality, consistency, and packaging. For the deployment phase, they face another set of challenges, ranging from custom requirements for data pre- and postprocessing, inconsistencies across frameworks, and lack of standardization in serving APIs. The goal of the IBM Developer Model Asset eXchange (MAX) is to remove these barriers to entry for developers to obtain, train, and deploy open source deep learning models for their business applications. In building the exchange, we encountered all these challenges and more. For the training phase, we leverage the Fabric for Deep Learning (FfDL), an open source project providing framework-independent training of deep learning models on Kubernetes. For the deployment phase, MAX provides standardized container-based, fully self-contained model artifacts encompassing the end-to-end deep learning predictive pipeline. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/max-aiconf-oct18-np-181011124309-thumbnail.jpg?width=120&height=120&fit=bounds" /> The common perception of applying deep learning is that you take an open source or research model, train it on raw data, and deploy the result as a fully self-contained artefact. The reality is far more complex. For the training phase, users face an array of challenges including handling varied deep learning frameworks, hardware requirements and configurations, not to mention code quality, consistency, and packaging. For the deployment phase, they face another set of challenges, ranging from custom requirements for data pre- and postprocessing, inconsistencies across frameworks, and lack of standardization in serving APIs. The goal of the IBM Developer Model Asset eXchange (MAX) is to remove these barriers to entry for developers to obtain, train, and deploy open source deep learning models for their business applications. In building the exchange, we encountered all these challenges and more. For the training phase, we leverage the Fabric for Deep Learning (FfDL), an open source project providing framework-independent training of deep learning models on Kubernetes. For the deployment phase, MAX provides standardized container-based, fully self-contained model artifacts encompassing the end-to-end deep learning predictive pipeline.

IBM Developer Model Asset eXchange from Nick Pentreath

]]> 538 3 https://cdn.slidesharecdn.com/ss_thumbnails/max-aiconf-oct18-np-181011124309-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 IBM Developer Model Asset eXchange - Deep Learning for Everyone /slideshow/ibm-developer-model-asset-exchange-deep-learning-for-everyone/118977601 max-aiconfmeetup-oct18-np-181010094358
We’ve all heard that AI is going to become as ubiquitous in the enterprise as the telephone, but what does that mean exactly?Everyone in a company has a telephone; and everyone knows how to use their telephone; and yet the company isn’t a phone company. How do we bring AI to the same standard of ubiquity —where everyone in a company has access to AI and knows how to use AI; and yet the company is not an AI company? In this talk, we’ll break down the challenges a domain expert faces today in applying AI to real-world problems. We’ll talk about the challenges that a domain expert needs to overcome in order to go from “I know a model of this type exists”to “I can tell an application developer how to apply this model to my domain.” We’ll conclude the talk with a live demo that show cases how a domain expert can cut through the stages of model deployment in minutes instead of days using the IBM Developer Model Asset Exchange.]]>
We’ve all heard that AI is going to become as ubiquitous in the enterprise as the telephone, but what does that mean exactly?Everyone in a company has a telephone; and everyone knows how to use their telephone; and yet the company isn’t a phone company. How do we bring AI to the same standard of ubiquity —where everyone in a company has access to AI and knows how to use AI; and yet the company is not an AI company? In this talk, we’ll break down the challenges a domain expert faces today in applying AI to real-world problems. We’ll talk about the challenges that a domain expert needs to overcome in order to go from “I know a model of this type exists”to “I can tell an application developer how to apply this model to my domain.” We’ll conclude the talk with a live demo that show cases how a domain expert can cut through the stages of model deployment in minutes instead of days using the IBM Developer Model Asset Exchange.]]> Wed, 10 Oct 2018 09:43:58 GMT /slideshow/ibm-developer-model-asset-exchange-deep-learning-for-everyone/118977601 NickPentreath@slideshare.net(NickPentreath) IBM Developer Model Asset eXchange - Deep Learning for Everyone NickPentreath We’ve all heard that AI is going to become as ubiquitous in the enterprise as the telephone, but what does that mean exactly?Everyone in a company has a telephone; and everyone knows how to use their telephone; and yet the company isn’t a phone company. How do we bring AI to the same standard of ubiquity —where everyone in a company has access to AI and knows how to use AI; and yet the company is not an AI company? In this talk, we’ll break down the challenges a domain expert faces today in applying AI to real-world problems. We’ll talk about the challenges that a domain expert needs to overcome in order to go from “I know a model of this type exists”to “I can tell an application developer how to apply this model to my domain.” We’ll conclude the talk with a live demo that show cases how a domain expert can cut through the stages of model deployment in minutes instead of days using the IBM Developer Model Asset Exchange. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/max-aiconfmeetup-oct18-np-181010094358-thumbnail.jpg?width=120&height=120&fit=bounds" /> We’ve all heard that AI is going to become as ubiquitous in the enterprise as the telephone, but what does that mean exactly?Everyone in a company has a telephone; and everyone knows how to use their telephone; and yet the company isn’t a phone company. How do we bring AI to the same standard of ubiquity —where everyone in a company has access to AI and knows how to use AI; and yet the company is not an AI company? In this talk, we’ll break down the challenges a domain expert faces today in applying AI to real-world problems. We’ll talk about the challenges that a domain expert needs to overcome in order to go from “I know a model of this type exists”to “I can tell an application developer how to apply this model to my domain.” We’ll conclude the talk with a live demo that show cases how a domain expert can cut through the stages of model deployment in minutes instead of days using the IBM Developer Model Asset Exchange.

IBM Developer Model Asset eXchange - Deep Learning for Everyone from Nick Pentreath

]]> 135 5 https://cdn.slidesharecdn.com/ss_thumbnails/max-aiconfmeetup-oct18-np-181010094358-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Search and Recommendations: 3 Sides of the Same Coin /slideshow/search-and-recommendations-3-sides-of-the-same-coin/102275713 searchrec-bbuzz-jun18-np-180613082300
Recommendation engines are one of the most well-known, widely-used and highest value use cases for applied machine learning. Search and recommender systems are closely linked, often co-existing and intermingling. Indeed, modern search applications at scale typically involve significant elements of machine learning, while personalization systems rely heavily on and are deeply integrated with search engines. In this session, I will explore this link between search and recommendations. In particular, I will cover three of the most common approaches for using search engines to serve personalized recommendation models. I call these the "score then search", "native search" and "custom ranking" approaches. I will detail each approach, comparing it with the others in terms of various considerations important for production systems at scale, including the architecture, schemas, performance, quality and flexibility aspects. Finally, I will also contrast these model-based approaches with what is achievable using pure search.]]>
Recommendation engines are one of the most well-known, widely-used and highest value use cases for applied machine learning. Search and recommender systems are closely linked, often co-existing and intermingling. Indeed, modern search applications at scale typically involve significant elements of machine learning, while personalization systems rely heavily on and are deeply integrated with search engines. In this session, I will explore this link between search and recommendations. In particular, I will cover three of the most common approaches for using search engines to serve personalized recommendation models. I call these the "score then search", "native search" and "custom ranking" approaches. I will detail each approach, comparing it with the others in terms of various considerations important for production systems at scale, including the architecture, schemas, performance, quality and flexibility aspects. Finally, I will also contrast these model-based approaches with what is achievable using pure search.]]> Wed, 13 Jun 2018 08:23:00 GMT /slideshow/search-and-recommendations-3-sides-of-the-same-coin/102275713 NickPentreath@slideshare.net(NickPentreath) Search and Recommendations: 3 Sides of the Same Coin NickPentreath Recommendation engines are one of the most well-known, widely-used and highest value use cases for applied machine learning. Search and recommender systems are closely linked, often co-existing and intermingling. Indeed, modern search applications at scale typically involve significant elements of machine learning, while personalization systems rely heavily on and are deeply integrated with search engines. In this session, I will explore this link between search and recommendations. In particular, I will cover three of the most common approaches for using search engines to serve personalized recommendation models. I call these the "score then search", "native search" and "custom ranking" approaches. I will detail each approach, comparing it with the others in terms of various considerations important for production systems at scale, including the architecture, schemas, performance, quality and flexibility aspects. Finally, I will also contrast these model-based approaches with what is achievable using pure search. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/searchrec-bbuzz-jun18-np-180613082300-thumbnail.jpg?width=120&height=120&fit=bounds" /> Recommendation engines are one of the most well-known, widely-used and highest value use cases for applied machine learning. Search and recommender systems are closely linked, often co-existing and intermingling. Indeed, modern search applications at scale typically involve significant elements of machine learning, while personalization systems rely heavily on and are deeply integrated with search engines. In this session, I will explore this link between search and recommendations. In particular, I will cover three of the most common approaches for using search engines to serve personalized recommendation models. I call these the "score then search", "native search" and "custom ranking" approaches. I will detail each approach, comparing it with the others in terms of various considerations important for production systems at scale, including the architecture, schemas, performance, quality and flexibility aspects. Finally, I will also contrast these model-based approaches with what is achievable using pure search.

Search and Recommendations: 3 Sides of the Same Coin from Nick Pentreath

]]> 230 3 https://cdn.slidesharecdn.com/ss_thumbnails/searchrec-bbuzz-jun18-np-180613082300-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Deep Learning for Recommender Systems /slideshow/deep-learning-for-recommender-systems-101178071/101178071 dlrs-sparksummit-jun18-np-180607180315
In the last few years, deep learning has achieved significant success in a wide range of domains, including computer vision, artificial intelligence, speech, NLP, and reinforcement learning. However, deep learning in recommender systems has, until recently, received relatively little attention. This talks explores recent advances in this area in both research and practice. I will explain how deep learning can be applied to recommendation settings, architectures for handling contextual data, side information, and time-based models.]]>
In the last few years, deep learning has achieved significant success in a wide range of domains, including computer vision, artificial intelligence, speech, NLP, and reinforcement learning. However, deep learning in recommender systems has, until recently, received relatively little attention. This talks explores recent advances in this area in both research and practice. I will explain how deep learning can be applied to recommendation settings, architectures for handling contextual data, side information, and time-based models.]]> Thu, 07 Jun 2018 18:03:15 GMT /slideshow/deep-learning-for-recommender-systems-101178071/101178071 NickPentreath@slideshare.net(NickPentreath) Deep Learning for Recommender Systems NickPentreath In the last few years, deep learning has achieved significant success in a wide range of domains, including computer vision, artificial intelligence, speech, NLP, and reinforcement learning. However, deep learning in recommender systems has, until recently, received relatively little attention. This talks explores recent advances in this area in both research and practice. I will explain how deep learning can be applied to recommendation settings, architectures for handling contextual data, side information, and time-based models. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/dlrs-sparksummit-jun18-np-180607180315-thumbnail.jpg?width=120&height=120&fit=bounds" /> In the last few years, deep learning has achieved significant success in a wide range of domains, including computer vision, artificial intelligence, speech, NLP, and reinforcement learning. However, deep learning in recommender systems has, until recently, received relatively little attention. This talks explores recent advances in this area in both research and practice. I will explain how deep learning can be applied to recommendation settings, architectures for handling contextual data, side information, and time-based models.

Deep Learning for Recommender Systems from Nick Pentreath

]]> 463 8 https://cdn.slidesharecdn.com/ss_thumbnails/dlrs-sparksummit-jun18-np-180607180315-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 Productionizing Spark ML Pipelines with the Portable Format for Analytics /slideshow/productionizing-spark-ml-pipelines-with-the-portable-format-for-analytics-100788521/100788521 sparkmlpfa-sparksummit-jun-18-np-180605204541
The common perception of machine learning is that it starts with data and ends with a model. In real-world production systems, the traditional data science and machine learning workflow of data preparation, feature engineering and model selection, while important, is only one aspect. A critical missing piece is the deployment and management of models, as well as the integration between the model creation and deployment phases. This is particularly challenging in the case of deploying Apache Spark ML pipelines for low-latency scoring. While MLlib’s DataFrame API is powerful and elegant, it is relatively ill-suited to the needs of many real-time predictive applications, in part because it is tightly coupled with the Spark SQL runtime. In this talk I will introduce the Portable Format for Analytics (PFA) for portable, open and standardized deployment of data science pipelines & analytic applications. I’ll also introduce and evaluate Aardpfark, a library for exporting Spark ML pipelines to PFA.]]>
The common perception of machine learning is that it starts with data and ends with a model. In real-world production systems, the traditional data science and machine learning workflow of data preparation, feature engineering and model selection, while important, is only one aspect. A critical missing piece is the deployment and management of models, as well as the integration between the model creation and deployment phases. This is particularly challenging in the case of deploying Apache Spark ML pipelines for low-latency scoring. While MLlib’s DataFrame API is powerful and elegant, it is relatively ill-suited to the needs of many real-time predictive applications, in part because it is tightly coupled with the Spark SQL runtime. In this talk I will introduce the Portable Format for Analytics (PFA) for portable, open and standardized deployment of data science pipelines & analytic applications. I’ll also introduce and evaluate Aardpfark, a library for exporting Spark ML pipelines to PFA.]]> Tue, 05 Jun 2018 20:45:41 GMT /slideshow/productionizing-spark-ml-pipelines-with-the-portable-format-for-analytics-100788521/100788521 NickPentreath@slideshare.net(NickPentreath) Productionizing Spark ML Pipelines with the Portable Format for Analytics NickPentreath The common perception of machine learning is that it starts with data and ends with a model. In real-world production systems, the traditional data science and machine learning workflow of data preparation, feature engineering and model selection, while important, is only one aspect. A critical missing piece is the deployment and management of models, as well as the integration between the model creation and deployment phases. This is particularly challenging in the case of deploying Apache Spark ML pipelines for low-latency scoring. While MLlib’s DataFrame API is powerful and elegant, it is relatively ill-suited to the needs of many real-time predictive applications, in part because it is tightly coupled with the Spark SQL runtime. In this talk I will introduce the Portable Format for Analytics (PFA) for portable, open and standardized deployment of data science pipelines & analytic applications. I’ll also introduce and evaluate Aardpfark, a library for exporting Spark ML pipelines to PFA. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/sparkmlpfa-sparksummit-jun-18-np-180605204541-thumbnail.jpg?width=120&height=120&fit=bounds" /> The common perception of machine learning is that it starts with data and ends with a model. In real-world production systems, the traditional data science and machine learning workflow of data preparation, feature engineering and model selection, while important, is only one aspect. A critical missing piece is the deployment and management of models, as well as the integration between the model creation and deployment phases. This is particularly challenging in the case of deploying Apache Spark ML pipelines for low-latency scoring. While MLlib’s DataFrame API is powerful and elegant, it is relatively ill-suited to the needs of many real-time predictive applications, in part because it is tightly coupled with the Spark SQL runtime. In this talk I will introduce the Portable Format for Analytics (PFA) for portable, open and standardized deployment of data science pipelines & analytic applications. I’ll also introduce and evaluate Aardpfark, a library for exporting Spark ML pipelines to PFA.

Productionizing Spark ML Pipelines with the Portable Format for Analytics from Nick Pentreath

]]> 990 4 https://cdn.slidesharecdn.com/ss_thumbnails/sparkmlpfa-sparksummit-jun-18-np-180605204541-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 RNNs for Recommendations and Personalization /slideshow/rnns-for-recommendations-and-personalization/95736134 rnn-recsys-strataai-may18-npentreath-180502181411
In the last few years, RNNs have achieved significant success in modeling time series and sequence data, in particular within the speech, language, and text domains. Recently, these techniques have been begun to be applied to session-based recommendation tasks, with very promising results. Nick Pentreath explores the latest research advances in this domain, as well as practical applications.]]>
In the last few years, RNNs have achieved significant success in modeling time series and sequence data, in particular within the speech, language, and text domains. Recently, these techniques have been begun to be applied to session-based recommendation tasks, with very promising results. Nick Pentreath explores the latest research advances in this domain, as well as practical applications.]]> Wed, 02 May 2018 18:14:10 GMT /slideshow/rnns-for-recommendations-and-personalization/95736134 NickPentreath@slideshare.net(NickPentreath) RNNs for Recommendations and Personalization NickPentreath In the last few years, RNNs have achieved significant success in modeling time series and sequence data, in particular within the speech, language, and text domains. Recently, these techniques have been begun to be applied to session-based recommendation tasks, with very promising results. Nick Pentreath explores the latest research advances in this domain, as well as practical applications. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/rnn-recsys-strataai-may18-npentreath-180502181411-thumbnail.jpg?width=120&height=120&fit=bounds" /> In the last few years, RNNs have achieved significant success in modeling time series and sequence data, in particular within the speech, language, and text domains. Recently, these techniques have been begun to be applied to session-based recommendation tasks, with very promising results. Nick Pentreath explores the latest research advances in this domain, as well as practical applications.

RNNs for Recommendations and Personalization from Nick Pentreath

]]> 942 4 https://cdn.slidesharecdn.com/ss_thumbnails/rnn-recsys-strataai-may18-npentreath-180502181411-thumbnail.jpg?width=120&height=120&fit=bounds presentation Black http://activitystrea.ms/schema/1.0/post

http://activitystrea.ms/schema/1.0/posted

0 https://public.slidesharecdn.com/v2/images/profile-picture.png I am a principal engineer at IBM working primarily on machine learning on Apache Spark. I am an Apache Spark committer and PMC member and author of Machine Learning with Spark. Previously, I cofounded Graphflow, a machine learning startup focused on recommendations. I've also worked at Goldman Sachs, Cognitive Match, and Mxit. I am passionate about combining commercial focus with machine learning and cutting-edge technology to build intelligent systems that learn from data to add business value. ibm.com https://cdn.slidesharecdn.com/ss_thumbnails/dataaisummiteu-2020-np-final-201118145214-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/notebookbased-ai-pipelines-with-elyra-and-kubeflow/239318441 Notebook-based AI Pipe... https://cdn.slidesharecdn.com/ss_thumbnails/sparkaisummit-2020-np-200624190925-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/scaling-up-deep-learning-by-scaling-down/236164597 Scaling up deep learni... https://cdn.slidesharecdn.com/ss_thumbnails/onnxpipelines-deployai-mar2020-np-200601125812-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/endtoend-deep-learning-deployment-with-onnx/234823747 End-to-End Deep Learni...