�ݺ�ߣshows by User: PierreGutierrez2

�ݺ�ߣshows by User: PierreGutierrez2 / http://www.slideshare.net/images/logo.gif �ݺ�ߣshows by User: PierreGutierrez2 / Fri, 12 May 2017 11:04:14 GMT �ݺ�ߣShare feed for �ݺ�ߣshows by User: PierreGutierrez2 Pragmatic deep learning for image labelling /PierreGutierrez2/pres-min pres-min-170512110414
This talk is composed of 3 major parts: the iterative creation of a recommender engine, the labeling of images, the post processing of images. After introducing the main topic, labeling images to improve recommendation engine performances, we start with a recommendation engine discussion. We briefly describe the “classical” recommender system (collaborative filtering, content based filtering) and their advantages and limitations. We then describe the re-ranking approach we used to combine different engines into one. Re-ranking is a method (used by Google for example) that takes the different ranking as features and optimizes a certain loss. In our case we combine our different recommendations through a logistic regression that predict the probability of purchases for each tuple (user, sale). This version of the engine led to +7% revenue per customer and is now running in production. We then explain why we wanted to use images information. It seemed that sales with some given images were performing better than others. If we had labels on all images we could use them in a content-based recommender system (used itself in the re-ranking engine). We then described how to label our images using pre-trained models, transfer learning and external APIs. We also show how easy it is to steal these APIs. The final part deals with post processing of the images. Since most pre-trained models only output one class prediction, we need to reshape these into broad themes that can be used in our engine. We use a Non Negative Matrix Factorization for this purpose and show that we have very interpretable results. We conclude by comparing visually the different engines. The key take away (more information in the pitch part) are theses: - Machine learning: overview of recommender systems, re-ranking, how to label images, transfer learning. - Do iterative data science. Start simple, then try more complex systems. - Avoid rushing in deep learning without checking what you can find on Internet. Use pre-trained models and transfer learning. There is a lot of hype around deep learning and image recognition. However, there are not that many success stories for web pure player companies. In our case we explain how we started with simple recommender systems before improving them gradually and finally using images information. One of the key take away is the following: do iterative data science. Always prefer shipping a minimum viable product before creating something complex. At our clients, we commonly see teams rushing into images projects for the only purpose of doing deep learning without a clear ROI in mind. We insist on the fact that deep learning is not an end in itself. Here, it boils down to making new information available in the system. In this sense, deep learning methods are just an extension of Business Intelligence. ]]>
This talk is composed of 3 major parts: the iterative creation of a recommender engine, the labeling of images, the post processing of images. After introducing the main topic, labeling images to improve recommendation engine performances, we start with a recommendation engine discussion. We briefly describe the “classical” recommender system (collaborative filtering, content based filtering) and their advantages and limitations. We then describe the re-ranking approach we used to combine different engines into one. Re-ranking is a method (used by Google for example) that takes the different ranking as features and optimizes a certain loss. In our case we combine our different recommendations through a logistic regression that predict the probability of purchases for each tuple (user, sale). This version of the engine led to +7% revenue per customer and is now running in production. We then explain why we wanted to use images information. It seemed that sales with some given images were performing better than others. If we had labels on all images we could use them in a content-based recommender system (used itself in the re-ranking engine). We then described how to label our images using pre-trained models, transfer learning and external APIs. We also show how easy it is to steal these APIs. The final part deals with post processing of the images. Since most pre-trained models only output one class prediction, we need to reshape these into broad themes that can be used in our engine. We use a Non Negative Matrix Factorization for this purpose and show that we have very interpretable results. We conclude by comparing visually the different engines. The key take away (more information in the pitch part) are theses: - Machine learning: overview of recommender systems, re-ranking, how to label images, transfer learning. - Do iterative data science. Start simple, then try more complex systems. - Avoid rushing in deep learning without checking what you can find on Internet. Use pre-trained models and transfer learning. There is a lot of hype around deep learning and image recognition. However, there are not that many success stories for web pure player companies. In our case we explain how we started with simple recommender systems before improving them gradually and finally using images information. One of the key take away is the following: do iterative data science. Always prefer shipping a minimum viable product before creating something complex. At our clients, we commonly see teams rushing into images projects for the only purpose of doing deep learning without a clear ROI in mind. We insist on the fact that deep learning is not an end in itself. Here, it boils down to making new information available in the system. In this sense, deep learning methods are just an extension of Business Intelligence. ]]> Fri, 12 May 2017 11:04:14 GMT /PierreGutierrez2/pres-min PierreGutierrez2@slideshare.net(PierreGutierrez2) Pragmatic deep learning for image labelling PierreGutierrez2 This talk is composed of 3 major parts: the iterative creation of a recommender engine, the labeling of images, the post processing of images. After introducing the main topic, labeling images to improve recommendation engine performances, we start with a recommendation engine discussion. We briefly describe the “classical” recommender system (collaborative filtering, content based filtering) and their advantages and limitations. We then describe the re-ranking approach we used to combine different engines into one. Re-ranking is a method (used by Google for example) that takes the different ranking as features and optimizes a certain loss. In our case we combine our different recommendations through a logistic regression that predict the probability of purchases for each tuple (user, sale). This version of the engine led to +7% revenue per customer and is now running in production. We then explain why we wanted to use images information. It seemed that sales with some given images were performing better than others. If we had labels on all images we could use them in a content-based recommender system (used itself in the re-ranking engine). We then described how to label our images using pre-trained models, transfer learning and external APIs. We also show how easy it is to steal these APIs. The final part deals with post processing of the images. Since most pre-trained models only output one class prediction, we need to reshape these into broad themes that can be used in our engine. We use a Non Negative Matrix Factorization for this purpose and show that we have very interpretable results. We conclude by comparing visually the different engines. The key take away (more information in the pitch part) are theses: - Machine learning: overview of recommender systems, re-ranking, how to label images, transfer learning. - Do iterative data science. Start simple, then try more complex systems. - Avoid rushing in deep learning without checking what you can find on Internet. Use pre-trained models and transfer learning. There is a lot of hype around deep learning and image recognition. However, there are not that many success stories for web pure player companies. In our case we explain how we started with simple recommender systems before improving them gradually and finally using images information. One of the key take away is the following: do iterative data science. Always prefer shipping a minimum viable product before creating something complex. At our clients, we commonly see teams rushing into images projects for the only purpose of doing deep learning without a clear ROI in mind. We insist on the fact that deep learning is not an end in itself. Here, it boils down to making new information available in the system. In this sense, deep learning methods are just an extension of Business Intelligence. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/pres-min-170512110414-thumbnail.jpg?width=120&height=120&fit=bounds" /> This talk is composed of 3 major parts: the iterative creation of a recommender engine, the labeling of images, the post processing of images. After introducing the main topic, labeling images to improve recommendation engine performances, we start with a recommendation engine discussion. We briefly describe the “classical” recommender system (collaborative filtering, content based filtering) and their advantages and limitations. We then describe the re-ranking approach we used to combine different engines into one. Re-ranking is a method (used by Google for example) that takes the different ranking as features and optimizes a certain loss. In our case we combine our different recommendations through a logistic regression that predict the probability of purchases for each tuple (user, sale). This version of the engine led to +7% revenue per customer and is now running in production. We then explain why we wanted to use images information. It seemed that sales with some given images were performing better than others. If we had labels on all images we could use them in a content-based recommender system (used itself in the re-ranking engine). We then described how to label our images using pre-trained models, transfer learning and external APIs. We also show how easy it is to steal these APIs. The final part deals with post processing of the images. Since most pre-trained models only output one class prediction, we need to reshape these into broad themes that can be used in our engine. We use a Non Negative Matrix Factorization for this purpose and show that we have very interpretable results. We conclude by comparing visually the different engines. The key take away (more information in the pitch part) are theses: - Machine learning: overview of recommender systems, re-ranking, how to label images, transfer learning. - Do iterative data science. Start simple, then try more complex systems. - Avoid rushing in deep learning without checking what you can find on Internet. Use pre-trained models and transfer learning. There is a lot of hype around deep learning and image recognition. However, there are not that many success stories for web pure player companies. In our case we explain how we started with simple recommender systems before improving them gradually and finally using images information. One of the key take away is the following: do iterative data science. Always prefer shipping a minimum viable product before creating something complex. At our clients, we commonly see teams rushing into images projects for the only purpose of doing deep learning without a clear ROI in mind. We insist on the fact that deep learning is not an end in itself. Here, it boils down to making new information available in the system. In this sense, deep learning methods are just an extension of Business Intelligence.

Pragmatic deep learning for image labelling from Pierre Gutierrez

]]> 1663 9 https://cdn.slidesharecdn.com/ss_thumbnails/pres-min-170512110414-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 From Labelling Open data images to building a private recommender system /slideshow/from-labelling-open-data-images-to-building-a-private-recommender-system/71144951 odscpres-170118140803
Recommender systems are paramount for e-business companies. There is an increasing need to take into account all the user information to tailor the best product proposition. One of them is the content that the user actually sees: the visual of the product. When it comes to hostels, some people can be more attracted by pictures of the room, the building or even the nearby beach. In this talk, we will describe how we improved an e-business vacation retailer recommender system using the content of images. We’ll explain how to leverage open dataset and pre-trained deep learning models to derive user taste information. This transfer learning approach enables companies to use state of the art machine learning methods without having deep learning expertise. ]]>
Recommender systems are paramount for e-business companies. There is an increasing need to take into account all the user information to tailor the best product proposition. One of them is the content that the user actually sees: the visual of the product. When it comes to hostels, some people can be more attracted by pictures of the room, the building or even the nearby beach. In this talk, we will describe how we improved an e-business vacation retailer recommender system using the content of images. We’ll explain how to leverage open dataset and pre-trained deep learning models to derive user taste information. This transfer learning approach enables companies to use state of the art machine learning methods without having deep learning expertise. ]]> Wed, 18 Jan 2017 14:08:03 GMT /slideshow/from-labelling-open-data-images-to-building-a-private-recommender-system/71144951 PierreGutierrez2@slideshare.net(PierreGutierrez2) From Labelling Open data images to building a private recommender system PierreGutierrez2 Recommender systems are paramount for e-business companies. There is an increasing need to take into account all the user information to tailor the best product proposition. One of them is the content that the user actually sees: the visual of the product. When it comes to hostels, some people can be more attracted by pictures of the room, the building or even the nearby beach. In this talk, we will describe how we improved an e-business vacation retailer recommender system using the content of images. We’ll explain how to leverage open dataset and pre-trained deep learning models to derive user taste information. This transfer learning approach enables companies to use state of the art machine learning methods without having deep learning expertise. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/odscpres-170118140803-thumbnail.jpg?width=120&height=120&fit=bounds" /> Recommender systems are paramount for e-business companies. There is an increasing need to take into account all the user information to tailor the best product proposition. One of them is the content that the user actually sees: the visual of the product. When it comes to hostels, some people can be more attracted by pictures of the room, the building or even the nearby beach. In this talk, we will describe how we improved an e-business vacation retailer recommender system using the content of images. We’ll explain how to leverage open dataset and pre-trained deep learning models to derive user taste information. This transfer learning approach enables companies to use state of the art machine learning methods without having deep learning expertise.

From Labelling Open data images to building a private recommender system from Pierre Gutierrez

]]> 1961 7 https://cdn.slidesharecdn.com/ss_thumbnails/odscpres-170118140803-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post

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

0 Machine learning and Internet of Things, the future of medical prevention /slideshow/machine-learning-and-internet-of-things-the-future-of-medical-prevention/67923442 datanativespres-161031143514
Title: "Machine learning and Internet of Things, the future of medical prevention" Abstract: In this talk, Pierre Gutierrez, a data scientist at Dataiku, will discuss Dataiku's experiences using machine learning on IOT data. We will talk about the challenges processing and cleaning IoT data, and how to successfully train a model that can be deployed in production. We will illustrate our talk with two examples from our previous work. Creating algorithm for early epilepsy seizure detection based on wearable tech and Detecting people activity through sensor data.]]>
Title: "Machine learning and Internet of Things, the future of medical prevention" Abstract: In this talk, Pierre Gutierrez, a data scientist at Dataiku, will discuss Dataiku's experiences using machine learning on IOT data. We will talk about the challenges processing and cleaning IoT data, and how to successfully train a model that can be deployed in production. We will illustrate our talk with two examples from our previous work. Creating algorithm for early epilepsy seizure detection based on wearable tech and Detecting people activity through sensor data.]]> Mon, 31 Oct 2016 14:35:14 GMT /slideshow/machine-learning-and-internet-of-things-the-future-of-medical-prevention/67923442 PierreGutierrez2@slideshare.net(PierreGutierrez2) Machine learning and Internet of Things, the future of medical prevention PierreGutierrez2 Title: "Machine learning and Internet of Things, the future of medical prevention" Abstract: In this talk, Pierre Gutierrez, a data scientist at Dataiku, will discuss Dataiku's experiences using machine learning on IOT data. We will talk about the challenges processing and cleaning IoT data, and how to successfully train a model that can be deployed in production. We will illustrate our talk with two examples from our previous work. Creating algorithm for early epilepsy seizure detection based on wearable tech and Detecting people activity through sensor data. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/datanativespres-161031143514-thumbnail.jpg?width=120&height=120&fit=bounds" /> Title: "Machine learning and Internet of Things, the future of medical prevention" Abstract: In this talk, Pierre Gutierrez, a data scientist at Dataiku, will discuss Dataiku's experiences using machine learning on IOT data. We will talk about the challenges processing and cleaning IoT data, and how to successfully train a model that can be deployed in production. We will illustrate our talk with two examples from our previous work. Creating algorithm for early epilepsy seizure detection based on wearable tech and Detecting people activity through sensor data.

Machine learning and Internet of Things, the future of medical prevention from Pierre Gutierrez

]]> 1122 4 https://cdn.slidesharecdn.com/ss_thumbnails/datanativespres-161031143514-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post

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

0 Beyond Churn Prediction : An Introduction to uplift modeling /slideshow/beyond-churn-prediction-an-introduction-to-uplift-modeling/67132627 introductiontoupliftmodeling-pierregutierrezdataiku-161013152613
These slides are from a talk I at the papis conference in Boston in 2016. The main subject is uplift modelling. Starting from a churn model approach for an e-gaming company, we introduce when to apply uplift methods, how to mathematically model them, and finally, how to evaluate them. I tried to bridge the gap between causal inference theory and uplift theory, especially concerning how to properly cross validate the results. The notation used is the one from uplift modelling. ]]>
These slides are from a talk I at the papis conference in Boston in 2016. The main subject is uplift modelling. Starting from a churn model approach for an e-gaming company, we introduce when to apply uplift methods, how to mathematically model them, and finally, how to evaluate them. I tried to bridge the gap between causal inference theory and uplift theory, especially concerning how to properly cross validate the results. The notation used is the one from uplift modelling. ]]> Thu, 13 Oct 2016 15:26:13 GMT /slideshow/beyond-churn-prediction-an-introduction-to-uplift-modeling/67132627 PierreGutierrez2@slideshare.net(PierreGutierrez2) Beyond Churn Prediction : An Introduction to uplift modeling PierreGutierrez2 These slides are from a talk I at the papis conference in Boston in 2016. The main subject is uplift modelling. Starting from a churn model approach for an e-gaming company, we introduce when to apply uplift methods, how to mathematically model them, and finally, how to evaluate them. I tried to bridge the gap between causal inference theory and uplift theory, especially concerning how to properly cross validate the results. The notation used is the one from uplift modelling. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/introductiontoupliftmodeling-pierregutierrezdataiku-161013152613-thumbnail.jpg?width=120&height=120&fit=bounds" /> These slides are from a talk I at the papis conference in Boston in 2016. The main subject is uplift modelling. Starting from a churn model approach for an e-gaming company, we introduce when to apply uplift methods, how to mathematically model them, and finally, how to evaluate them. I tried to bridge the gap between causal inference theory and uplift theory, especially concerning how to properly cross validate the results. The notation used is the one from uplift modelling.

Beyond Churn Prediction : An Introduction to uplift modeling from Pierre Gutierrez

]]> 5584 9 https://cdn.slidesharecdn.com/ss_thumbnails/introductiontoupliftmodeling-pierregutierrezdataiku-161013152613-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post

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

0 Introduction to Uplift Modelling /slideshow/introduction-to-uplift-modelling/59011566 uplift-160303094318
These slides are from a talk I gave at Google Campus Madrid for the Machine Learning Meetup. The main subject is uplift modelling. Starting from a churn model approach for an e-gaming company, we introduce when to apply uplift methods, how to mathematically model them, and finally, how to evaluate them. ]]>
These slides are from a talk I gave at Google Campus Madrid for the Machine Learning Meetup. The main subject is uplift modelling. Starting from a churn model approach for an e-gaming company, we introduce when to apply uplift methods, how to mathematically model them, and finally, how to evaluate them. ]]> Thu, 03 Mar 2016 09:43:18 GMT /slideshow/introduction-to-uplift-modelling/59011566 PierreGutierrez2@slideshare.net(PierreGutierrez2) Introduction to Uplift Modelling PierreGutierrez2 These slides are from a talk I gave at Google Campus Madrid for the Machine Learning Meetup. The main subject is uplift modelling. Starting from a churn model approach for an e-gaming company, we introduce when to apply uplift methods, how to mathematically model them, and finally, how to evaluate them. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/uplift-160303094318-thumbnail.jpg?width=120&height=120&fit=bounds" /> These slides are from a talk I gave at Google Campus Madrid for the Machine Learning Meetup. The main subject is uplift modelling. Starting from a churn model approach for an e-gaming company, we introduce when to apply uplift methods, how to mathematically model them, and finally, how to evaluate them.

Introduction to Uplift Modelling from Pierre Gutierrez

]]> 12243 17 https://cdn.slidesharecdn.com/ss_thumbnails/uplift-160303094318-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 Before Kaggle /slideshow/before-kaggle/53202375 wiml-150925165635-lva1-app6892
Many think that a Data Science is like a Kaggle competition. There are, however big differences in the approach. This presentation is about designing carefully your evaluation scheme to avoid overfitting and unexpected production performances. ]]>
Many think that a Data Science is like a Kaggle competition. There are, however big differences in the approach. This presentation is about designing carefully your evaluation scheme to avoid overfitting and unexpected production performances. ]]> Fri, 25 Sep 2015 16:56:35 GMT /slideshow/before-kaggle/53202375 PierreGutierrez2@slideshare.net(PierreGutierrez2) Before Kaggle PierreGutierrez2 Many think that a Data Science is like a Kaggle competition. There are, however big differences in the approach. This presentation is about designing carefully your evaluation scheme to avoid overfitting and unexpected production performances. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/wiml-150925165635-lva1-app6892-thumbnail.jpg?width=120&height=120&fit=bounds" /> Many think that a Data Science is like a Kaggle competition. There are, however big differences in the approach. This presentation is about designing carefully your evaluation scheme to avoid overfitting and unexpected production performances.

Before Kaggle from Pierre Gutierrez

]]> 1952 5 https://cdn.slidesharecdn.com/ss_thumbnails/wiml-150925165635-lva1-app6892-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post

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

0 Churn prediction data modeling /slideshow/churn-prediction-data-modeling/51844480 churn-150820045051-lva1-app6892
�ݺ�ߣs from the presentation of this NYC meetup : http://www.meetup.com/Data-Modeling/events/224554990/ I talked about how to model churn before even thinking about the machine learning model.]]>
�ݺ�ߣs from the presentation of this NYC meetup : http://www.meetup.com/Data-Modeling/events/224554990/ I talked about how to model churn before even thinking about the machine learning model.]]> Thu, 20 Aug 2015 04:50:51 GMT /slideshow/churn-prediction-data-modeling/51844480 PierreGutierrez2@slideshare.net(PierreGutierrez2) Churn prediction data modeling PierreGutierrez2 �ݺ�ߣs from the presentation of this NYC meetup : http://www.meetup.com/Data-Modeling/events/224554990/ I talked about how to model churn before even thinking about the machine learning model. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/churn-150820045051-lva1-app6892-thumbnail.jpg?width=120&height=120&fit=bounds" /> �ݺ�ߣs from the presentation of this NYC meetup : http://www.meetup.com/Data-Modeling/events/224554990/ I talked about how to model churn before even thinking about the machine learning model.

Churn prediction data modeling from Pierre Gutierrez

]]> 4079 9 https://cdn.slidesharecdn.com/ss_thumbnails/churn-150820045051-lva1-app6892-thumbnail.jpg?width=120&height=120&fit=bounds presentation 000000 http://activitystrea.ms/schema/1.0/post

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

0 https://cdn.slidesharecdn.com/profile-photo-PierreGutierrez2-48x48.jpg?cb=1616861753 www.dataiku.com https://cdn.slidesharecdn.com/ss_thumbnails/pres-min-170512110414-thumbnail.jpg?width=320&height=320&fit=bounds PierreGutierrez2/pres-min Pragmatic deep learnin... https://cdn.slidesharecdn.com/ss_thumbnails/odscpres-170118140803-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/from-labelling-open-data-images-to-building-a-private-recommender-system/71144951 From Labelling Open da... https://cdn.slidesharecdn.com/ss_thumbnails/datanativespres-161031143514-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/machine-learning-and-internet-of-things-the-future-of-medical-prevention/67923442 Machine learning and I...