The Next Live Session: Exploring Distributed Machine Learning and Deep Learning Systems
Our next live session is a public live session that's accessible to everyone via our YouTube channel, and we're excited to share the topic with you. On the coming Saturday, we'll be discussing the architecture and design of distributed machine learning and deep learning systems.
The Live Session Details
------------------------
Please note that the live session will take place on Saturday, December 18th at 7:00 pm, not Sunday. We'll also provide a link to the live session in the description section below our YouTube channel. This live session is expected to be around 90 minutes long, during which we'll discuss various architectural choices and why they're sensible.
Prerequisites for the Live Session
-----------------------------------
While basic machine learning algorithms are recommended as a prerequisite, even if you know logistic regression or linear regression, you'll still be able to understand a good chunk of the discussion. However, if you have knowledge of deep learning techniques, particularly stochastic gradient descent and its variants, you'll be able to grasp 100% of the discussion and appreciate how to design these distributed systems from scratch.
The Focus on Architectural Choices
-----------------------------------
When we discuss architecture, we want to give you a little heads up that we'll be covering architectures used in top product-based companies like Google, Amazon, Netflix, and Uber. These are architectures that have been deployed based on research work, published technical blogs, and real-world experience. We'll also piggyback on examples from cloud platforms, such as AWS Sagemaker, Google Compute Platform, and Microsoft Azure.
Learning from Open Source Libraries
-------------------------------------
We believe in taking a lot of examples and ideas from open-source libraries to help you understand what's happening under the hood. For instance, we'll be looking at Spark ML, a very popular distributed machine learning platform, as well as TF Distributed and other examples available. If you take XGBoost, for example, there's even a distributed version that can work on a cluster of computers.
Design Patterns and Architectural Strategies
-------------------------------------------
We understand the design patterns and architectural strategies employed by these libraries because they're open-source and we can read the code to grasp what's happening under the hood. While we won't be diving into each library themselves, we'll be discussing the design choices and how you can deploy them in your own real-world systems.
Conclusion
----------
We're looking forward to this live session and excited to share our knowledge with you. We encourage you to share this with your friends who might benefit from it as well. See you soon!
"WEBVTTKind: captionsLanguage: enhi friends our next live session is a public live session that's accessible to everyone via our youtube channel and the topic we will discuss this time would be about the architecture and design of distributed machine learning and deep learning systems this live session will be on the coming saturday please note saturday not sunday which is the 18th of december at 7 00 pm on our youtube channel we will also provide a link to the live session on youtube in the description section box below again this live session will be roughly like a 90 minute live session where we will discuss the architecture various architectural choices we have why the why these architectural choices are sensible and where we should use what basic logic like that so the prerequisites for this whole live session basically how can you make the most out of it is if you know basic machine learning algorithms you can understand a good chunk of it even if you know basic logistic regression and linear regression you'll be able to get a sense of how distributed machine learning works but we will also cover some architectural choices that are specific to deep learning techniques because in a deep learning setup you have distributed gpus and distributed tpus typically so again underlying whole of logistic regression linear regression or even deep learning algorithms are basically stochastic gradient descent approaches and variants of it so if you know stochastic gradient descent that underlies your standard logistic and linear regression you can understand at least 80 90 percent of it if you know deep learning you'll be able to understand hundred percent of the discussion and you'll be able to appreciate and participate and understand how to design these distributed systems yourself from scratch cool so uh when we discuss about architecture just wanted to give you a little heads up there we will discuss about architectures that are actually used in the top product based companies companies like google amazon netflix uber these are architectures that they have deployed based on the research work they've published based on the technical blogs that they have published right we'll also piggyback and we'll also take some examples from cloud platforms for example how does aws sagemaker perform distributed training of large machine learning and deep learning systems similarly how does google compute platform have specialized tools to do that how does azure have that right so we'll also take some examples from the popular cloud platforms like aws gcp and microsoft azure most importantly we'll take lot of examples and a lot of ideas from open source libraries because you can see the code right so you understand exactly what's happening under the hood so we'll take spark ml which is a very very popular distributed machine learning platform we'll also take tf distributed and other examples that are available for example if you take xgboost there is a distributed version of xgboost which can work even on a cluster of computers so we'll talk about design choices we will not dive into each of these libraries themselves but we'll see what design patterns some of these libraries or what architectural design strategies these libraries have actually employed we understand very well about these strategies because they're open source we can read the code and understand what is happening under the hood right so cool so see you on the coming saturday we are looking forward for it it will roughly be a 90 minute session we'll try and cover a few examples of architectural choices and how you can deploy it in your own real-world systems again please note it's on the saturday not the sunday which is the 18th of december 7 pm the link to the live will be in the description box below we are thrilled to do this and looking forward to having all of you also please share this with your friends whom you think might benefit from this okay see you soon byehi friends our next live session is a public live session that's accessible to everyone via our youtube channel and the topic we will discuss this time would be about the architecture and design of distributed machine learning and deep learning systems this live session will be on the coming saturday please note saturday not sunday which is the 18th of december at 7 00 pm on our youtube channel we will also provide a link to the live session on youtube in the description section box below again this live session will be roughly like a 90 minute live session where we will discuss the architecture various architectural choices we have why the why these architectural choices are sensible and where we should use what basic logic like that so the prerequisites for this whole live session basically how can you make the most out of it is if you know basic machine learning algorithms you can understand a good chunk of it even if you know basic logistic regression and linear regression you'll be able to get a sense of how distributed machine learning works but we will also cover some architectural choices that are specific to deep learning techniques because in a deep learning setup you have distributed gpus and distributed tpus typically so again underlying whole of logistic regression linear regression or even deep learning algorithms are basically stochastic gradient descent approaches and variants of it so if you know stochastic gradient descent that underlies your standard logistic and linear regression you can understand at least 80 90 percent of it if you know deep learning you'll be able to understand hundred percent of the discussion and you'll be able to appreciate and participate and understand how to design these distributed systems yourself from scratch cool so uh when we discuss about architecture just wanted to give you a little heads up there we will discuss about architectures that are actually used in the top product based companies companies like google amazon netflix uber these are architectures that they have deployed based on the research work they've published based on the technical blogs that they have published right we'll also piggyback and we'll also take some examples from cloud platforms for example how does aws sagemaker perform distributed training of large machine learning and deep learning systems similarly how does google compute platform have specialized tools to do that how does azure have that right so we'll also take some examples from the popular cloud platforms like aws gcp and microsoft azure most importantly we'll take lot of examples and a lot of ideas from open source libraries because you can see the code right so you understand exactly what's happening under the hood so we'll take spark ml which is a very very popular distributed machine learning platform we'll also take tf distributed and other examples that are available for example if you take xgboost there is a distributed version of xgboost which can work even on a cluster of computers so we'll talk about design choices we will not dive into each of these libraries themselves but we'll see what design patterns some of these libraries or what architectural design strategies these libraries have actually employed we understand very well about these strategies because they're open source we can read the code and understand what is happening under the hood right so cool so see you on the coming saturday we are looking forward for it it will roughly be a 90 minute session we'll try and cover a few examples of architectural choices and how you can deploy it in your own real-world systems again please note it's on the saturday not the sunday which is the 18th of december 7 pm the link to the live will be in the description box below we are thrilled to do this and looking forward to having all of you also please share this with your friends whom you think might benefit from this okay see you soon bye\n"
