ºÝºÝߣshows by User: darian_f / http://www.slideshare.net/images/logo.gif ºÝºÝߣshows by User: darian_f / Sun, 07 Jul 2019 13:48:26 GMT ºÝºÝߣShare feed for ºÝºÝߣshows by User: darian_f Accelerating Deep Learning Inference 
on Mobile Systems /slideshow/accelerating-deep-learning-inference-on-mobile-systems/154112354 frajbergaims2019v2-190707134826
International Conference on AI and Mobile Services Services Conference Federation (SCF) San Diego, CA, USA June 2019 Artificial Intelligence on the edge is a matter of great importance towards the enhancement of smart devices that rely on operations with real-time constraints. Despite the rapid growth of computational power in embedded systems, such as smartphones, wearable devices, drones and FPGAs, the deployment of highly complex and considerably big models remains challenging. Optimized execution requires managing memory allocation efficiently, to avoid overloading, and exploiting the available hardware resources for acceleration, which is not trivial given the non standardized access to such resources. We present PolimiDL, an open source framework for the acceleration of Deep Learning inference on mobile and embedded systems with limited resources and heterogeneous architectures. Experimental results show competitive results w.r.t. TensorFlow Lite for the execution of small models.]]>
International Conference on AI and Mobile Services Services Conference Federation (SCF) San Diego, CA, USA June 2019 Artificial Intelligence on the edge is a matter of great importance towards the enhancement of smart devices that rely on operations with real-time constraints. Despite the rapid growth of computational power in embedded systems, such as smartphones, wearable devices, drones and FPGAs, the deployment of highly complex and considerably big models remains challenging. Optimized execution requires managing memory allocation efficiently, to avoid overloading, and exploiting the available hardware resources for acceleration, which is not trivial given the non standardized access to such resources. We present PolimiDL, an open source framework for the acceleration of Deep Learning inference on mobile and embedded systems with limited resources and heterogeneous architectures. Experimental results show competitive results w.r.t. TensorFlow Lite for the execution of small models.]]> Sun, 07 Jul 2019 13:48:26 GMT /slideshow/accelerating-deep-learning-inference-on-mobile-systems/154112354 darian_f@slideshare.net(darian_f) Accelerating Deep Learning Inference 
on Mobile Systems darian_f International Conference on AI and Mobile Services Services Conference Federation (SCF) San Diego, CA, USA June 2019 Artificial Intelligence on the edge is a matter of great importance towards the enhancement of smart devices that rely on operations with real-time constraints. Despite the rapid growth of computational power in embedded systems, such as smartphones, wearable devices, drones and FPGAs, the deployment of highly complex and considerably big models remains challenging. Optimized execution requires managing memory allocation efficiently, to avoid overloading, and exploiting the available hardware resources for acceleration, which is not trivial given the non standardized access to such resources. We present PolimiDL, an open source framework for the acceleration of Deep Learning inference on mobile and embedded systems with limited resources and heterogeneous architectures. Experimental results show competitive results w.r.t. TensorFlow Lite for the execution of small models. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/frajbergaims2019v2-190707134826-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> International Conference on AI and Mobile Services Services Conference Federation (SCF) San Diego, CA, USA June 2019 Artificial Intelligence on the edge is a matter of great importance towards the enhancement of smart devices that rely on operations with real-time constraints. Despite the rapid growth of computational power in embedded systems, such as smartphones, wearable devices, drones and FPGAs, the deployment of highly complex and considerably big models remains challenging. Optimized execution requires managing memory allocation efficiently, to avoid overloading, and exploiting the available hardware resources for acceleration, which is not trivial given the non standardized access to such resources. We present PolimiDL, an open source framework for the acceleration of Deep Learning inference on mobile and embedded systems with limited resources and heterogeneous architectures. Experimental results show competitive results w.r.t. TensorFlow Lite for the execution of small models.
Accelerating Deep Learning Inference on Mobile Systems from Darian Frajberg
]]> 318 3 https://cdn.slidesharecdn.com/ss_thumbnails/frajbergaims2019v2-190707134826-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 Applying Deep Learning with Weak and Noisy labels /slideshow/applying-deep-learning-with-weak-and-noisy-labels/115826357 weakandnoisylabelsdl-180921181328
Scientific seminar at Politecnico di Milano Como, Italy September 2018 In recent years, Deep Learning has achieved outstanding results outperforming previous techniques and even humans, thus becoming the state-of-the-art in a wide range of tasks, among which Computer Vision has been one of the most benefited areas. Nonetheless, most of this success is tightly coupled to strongly supervised learning tasks, which require highly accurate, expensive and labor-intensive defined ground truth labels. In this presentation, we will introduce diverse alternatives to deal with this problem and support the training of Deep Learning models for Computer Vision tasks by simplifying the process of data labelling or exploiting the unlimited supply of publicly available data in Internet (such as user-tagged images from Flickr). Such alternatives rely on data comprising noisy and weak labels, which are much easier to collect but require special care to be used.]]>
Scientific seminar at Politecnico di Milano Como, Italy September 2018 In recent years, Deep Learning has achieved outstanding results outperforming previous techniques and even humans, thus becoming the state-of-the-art in a wide range of tasks, among which Computer Vision has been one of the most benefited areas. Nonetheless, most of this success is tightly coupled to strongly supervised learning tasks, which require highly accurate, expensive and labor-intensive defined ground truth labels. In this presentation, we will introduce diverse alternatives to deal with this problem and support the training of Deep Learning models for Computer Vision tasks by simplifying the process of data labelling or exploiting the unlimited supply of publicly available data in Internet (such as user-tagged images from Flickr). Such alternatives rely on data comprising noisy and weak labels, which are much easier to collect but require special care to be used.]]> Fri, 21 Sep 2018 18:13:28 GMT /slideshow/applying-deep-learning-with-weak-and-noisy-labels/115826357 darian_f@slideshare.net(darian_f) Applying Deep Learning with Weak and Noisy labels darian_f Scientific seminar at Politecnico di Milano Como, Italy September 2018 In recent years, Deep Learning has achieved outstanding results outperforming previous techniques and even humans, thus becoming the state-of-the-art in a wide range of tasks, among which Computer Vision has been one of the most benefited areas. Nonetheless, most of this success is tightly coupled to strongly supervised learning tasks, which require highly accurate, expensive and labor-intensive defined ground truth labels. In this presentation, we will introduce diverse alternatives to deal with this problem and support the training of Deep Learning models for Computer Vision tasks by simplifying the process of data labelling or exploiting the unlimited supply of publicly available data in Internet (such as user-tagged images from Flickr). Such alternatives rely on data comprising noisy and weak labels, which are much easier to collect but require special care to be used. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/weakandnoisylabelsdl-180921181328-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> Scientific seminar at Politecnico di Milano Como, Italy September 2018 In recent years, Deep Learning has achieved outstanding results outperforming previous techniques and even humans, thus becoming the state-of-the-art in a wide range of tasks, among which Computer Vision has been one of the most benefited areas. Nonetheless, most of this success is tightly coupled to strongly supervised learning tasks, which require highly accurate, expensive and labor-intensive defined ground truth labels. In this presentation, we will introduce diverse alternatives to deal with this problem and support the training of Deep Learning models for Computer Vision tasks by simplifying the process of data labelling or exploiting the unlimited supply of publicly available data in Internet (such as user-tagged images from Flickr). Such alternatives rely on data comprising noisy and weak labels, which are much easier to collect but require special care to be used.
Applying Deep Learning with Weak and Noisy labels from Darian Frajberg
]]> 1080 4 https://cdn.slidesharecdn.com/ss_thumbnails/weakandnoisylabelsdl-180921181328-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 Introduction to the Artificial Intelligence and Computer Vision revolution /slideshow/introduction-to-the-artificial-intelligence-and-computer-vision-revolution/81372897 introductiontotheaiandcvrevolution-171030133024
Scientific seminar at Politecnico di Milano Como, Italy October 2017]]>
Scientific seminar at Politecnico di Milano Como, Italy October 2017]]> Mon, 30 Oct 2017 13:30:24 GMT /slideshow/introduction-to-the-artificial-intelligence-and-computer-vision-revolution/81372897 darian_f@slideshare.net(darian_f) Introduction to the Artificial Intelligence and Computer Vision revolution darian_f Scientific seminar at Politecnico di Milano Como, Italy October 2017 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/introductiontotheaiandcvrevolution-171030133024-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> Scientific seminar at Politecnico di Milano Como, Italy October 2017
Introduction to the Artificial Intelligence and Computer Vision revolution from Darian Frajberg
]]> 4191 5 https://cdn.slidesharecdn.com/ss_thumbnails/introductiontotheaiandcvrevolution-171030133024-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 Heterogeneous information integration for mountain augmented reality mobile apps /slideshow/heterogeneous-information-integration-for-mountain-augmented-reality-mobile-apps/81039793 frajbergdsaa2017-171021060446
DSAA2017 4th IEEE International Conference on Data Science and Advanced Analytics Tokyo, Japan October 2017]]>
DSAA2017 4th IEEE International Conference on Data Science and Advanced Analytics Tokyo, Japan October 2017]]> Sat, 21 Oct 2017 06:04:46 GMT /slideshow/heterogeneous-information-integration-for-mountain-augmented-reality-mobile-apps/81039793 darian_f@slideshare.net(darian_f) Heterogeneous information integration for mountain augmented reality mobile apps darian_f DSAA2017 4th IEEE International Conference on Data Science and Advanced Analytics Tokyo, Japan October 2017 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/frajbergdsaa2017-171021060446-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> DSAA2017 4th IEEE International Conference on Data Science and Advanced Analytics Tokyo, Japan October 2017
Heterogeneous information integration for mountain augmented reality mobile apps from Darian Frajberg
]]> 183 2 https://cdn.slidesharecdn.com/ss_thumbnails/frajbergdsaa2017-171021060446-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 Convolutional Neural Network for pixel-wise skyline detection /slideshow/convolutional-neural-network-for-pixelwise-skyline-detection-79815913/79815913 frajbergicann2017-170915165737
ICANN 2017 26th International Conference on Artificial Neural Networks Alghero, Sardinia, Italy September 2017]]>
ICANN 2017 26th International Conference on Artificial Neural Networks Alghero, Sardinia, Italy September 2017]]> Fri, 15 Sep 2017 16:57:37 GMT /slideshow/convolutional-neural-network-for-pixelwise-skyline-detection-79815913/79815913 darian_f@slideshare.net(darian_f) Convolutional Neural Network for pixel-wise skyline detection darian_f ICANN 2017 26th International Conference on Artificial Neural Networks Alghero, Sardinia, Italy September 2017 <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/frajbergicann2017-170915165737-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> ICANN 2017 26th International Conference on Artificial Neural Networks Alghero, Sardinia, Italy September 2017
Convolutional Neural Network for pixel-wise skyline detection from Darian Frajberg
]]> 446 5 https://cdn.slidesharecdn.com/ss_thumbnails/frajbergicann2017-170915165737-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 Volatile Functionality in Action: Methods, Techniques and Assessment /slideshow/volatile-functionality-in-action-methods-techniques-and-assessment/76963932 darian-frajberg-170615074333
16th International Conference on Web Engineering (ICWE2016), Lugano, Switzerland. June 2016.]]>
16th International Conference on Web Engineering (ICWE2016), Lugano, Switzerland. June 2016.]]> Thu, 15 Jun 2017 07:43:33 GMT /slideshow/volatile-functionality-in-action-methods-techniques-and-assessment/76963932 darian_f@slideshare.net(darian_f) Volatile Functionality in Action: Methods, Techniques and Assessment darian_f 16th International Conference on Web Engineering (ICWE2016), Lugano, Switzerland. June 2016. <img style="border:1px solid #C3E6D8;float:right;" alt="" src="https://cdn.slidesharecdn.com/ss_thumbnails/darian-frajberg-170615074333-thumbnail.jpg?width=120&amp;height=120&amp;fit=bounds" /><br> 16th International Conference on Web Engineering (ICWE2016), Lugano, Switzerland. June 2016.
Volatile Functionality in Action: Methods, Techniques and Assessment from Darian Frajberg
]]> 140 2 https://cdn.slidesharecdn.com/ss_thumbnails/darian-frajberg-170615074333-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://cdn.slidesharecdn.com/profile-photo-darian_f-48x48.jpg?cb=1562507246 Researcher at the Department of Electronics, Information and Bioengineering (DEIB), Politecnico di Milano, Italy. I was born in Argentina and I also have Polish citizenship. https://cdn.slidesharecdn.com/ss_thumbnails/frajbergaims2019v2-190707134826-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/accelerating-deep-learning-inference-on-mobile-systems/154112354 Accelerating Deep Lear... https://cdn.slidesharecdn.com/ss_thumbnails/weakandnoisylabelsdl-180921181328-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/applying-deep-learning-with-weak-and-noisy-labels/115826357 Applying Deep Learning... https://cdn.slidesharecdn.com/ss_thumbnails/introductiontotheaiandcvrevolution-171030133024-thumbnail.jpg?width=320&height=320&fit=bounds slideshow/introduction-to-the-artificial-intelligence-and-computer-vision-revolution/81372897 Introduction to the Ar...