NVIDIA Expert digital workshop series on building AI solutions

November 19, 2020

Building AI Solutions with NVIDIA

The following article is based on an online workshop conducted by the E2E - one of the largest cloud service providers in India; in collaboration with NVIDIA. The aim of this workshop is to serve as a guide for building AI solutions on NVIDIA accelerated computing platform. The primary agenda is to expound on upcoming AI-based avenues of Conversational AI, Recommender Systems, Healthcare and Lifesciences, Intelligent Video Analytics, etc.  The panel further discusses building AI solutions by leveraging NVIDIA NGC and NVIDIA A100 TENSOR CORE GPU over E2E cloud.

About the Panel 

The workshop has been moderated by Mr UnniKrishnan AR. He is the Head Developer, Ecosystem, at NVIDIA Deep Learning Institute and CSP Business, South Asia. The panel for the workshop also consists of Mr Ashish Sardana(Data Scientist, NVIDIA), Ms Pooja Venkatesh(Developer Relations Manager, IVA-NVIDIA) and Mr Pallab Maji(Senior Solutions Architect Deep Learning, NVIDIA).

Introduction to NVIDIA NGC Catalog

NVIDIA NGC is a software hub that contains the latest software tools designed to assist developers and infrastructure maintenance teams responsible for building an organisation’s software stack. Mr. Pallab gives an overview of the NVIDIA NGC as “a software hub that houses different artefacts required for algorithm development”. These artefacts available on NGC play a vital role in developing algorithms for machine learning, deep learning, and high-performance computing ecosystems. Developers can increase their efficiency multifold by availing all the best software development resources located at one place, rather than having to wander around the internet for the same. Interested individuals can register themselves here and browse through all of NGC’s offerings.

Conversational AI

“Language is a medium of communication and its use for human activities is obvious”, states Mr UnniKrishnan, as the workshop moves towards the topic of Conversational AI. NVIDIA has developed a full-fledged stack for supporting conversational AI that can be utilised by a wide array of users ranging from enterprises needing it for their internal application purposes to solution-providers looking to plug a conversational AI piece into their products. Mr Pallab furthers the discussion saying, “Currently, it is one of the best times to work on Conversational AI”. He backs his statement by explaining how there has been a multitude of innovations in speech recognition and language modelling technologies, both of which are pivotal in improving the accuracy of Conversational AI. The key here is to combine visual and speech cues accurately, and then to be able to deliver an appropriate response while maintaining the quality of service. The NVIDIA JARVIS module is a prime example of this, as it provides multi-modal conversational AI skills while delivering low-latency. Transcribing multi-user, multi-context conversations, and End-to-end conversational AI systems are two of many applications of the JARVIS model.

Recommender Systems

The moderator starts the next segment of the workshop with the following assertion, “Just like the Conversational AI, the most critical piece out there in the present age of AI is the personalisation of the Internet”. Mr Ashish Sardana explains how the recommenders will help make this possible. According to a report from McKinsey, “Already, 35% of what consumers purchase on Amazon and 75% of what they watch on Netflix come from product recommendations based on such algorithms”. It suggests that recommender systems help internet provider companies personalise the internet for users. Industries like E-commerce, digital content, etc. are heavily dependent on recommender systems to connect with their consumers. NVIDIA’s MERLIN is a framework for developing recommender systems that can be employed by developers for the needs of their respective organisations.

Healthcare and Lifesciences

Mr UnniKrishnan takes up the topic of Healthcare and Lifesciences for the penultimate section of the workshop. According to him, “The value proposition of AI and combination of AI with HPC towards the healthcare is phenomenal”. Mr Sardana covers this topic as well; elaborating the categorisation of CLARA framework into CLARA Guardian, CLARA Discovery and CLARA for Medical Imaging. Most of  NVIDIA’s work has been focused on the medical imaging aspect, which also happens to be the healthcare industry’s facet with most acceptance for AI. However, the present pandemic situation has put a great impetus on CLARA Guardian and CLARA Discovery as well. The former has tremendous applications in developing smart and contactless hospitals, which is the need of the hour. CLARA Discovery, on the other hand, is focussed on combinational genomics that is used for developing vaccines and drugs.

Intelligent Video Analytics

In the final segment of the workshop, Ms Pooja elaborates on the work NVIDIA has been doing in Intelligent Video Analytics. According to her, “The IVA market today is a two trillion dollar industry”. The purpose of implementing IVA is to gain predictive insights from the huge amount of video data that is being generated globally. These insights can be used across a plethora of industry segments like Traffic management, Warehouses & Logistics, Access control, etc. NVIDIA’s METROPOLIS stack is a software development kit which can be used at high levels of scalability across the above-mentioned industry applications.

The moderator conducted a brief Q&A session at the end to revisit all of the above topics, before wrapping things up. 

Latest Blogs
This is a decorative image for: What is SOTA in Artificial Intelligence?
August 5, 2022

What is SOTA in Artificial Intelligence?

If you are one of those people who love to pursue Artificial Intelligence and related operations like Machine Learning, then you must have certainly come across a term called SOTA. It is one of the much-talked things in the field of AI and holds a lot of gravity.

But for those who are interested yet are clueless about what SOTA is and what its relevance is in the field of AI, here is a simple definition of SOTA, what it means, and what importance it holds.  

What is SOTA?

SOTA is an acronym for State-Of-The-Art. In the context of Artificial Intelligence (AI), it refers to the best models that can be used for achieving the results in a task. Mind you; it should be an AI-specific task only. SOTA models can be applied in many ways in AI. It could either be applied to –

(a) Machine Learning (ML) tasks

(b) Deep Neural Networks (DNNs) tasks

(c) Natural Language Processing (NLPs) [this is a subset of deep neural networks]

(d) Generic tasks

How does SOTA help in AI?

Using SOTA models in AI has many benefits of its own. The primary benefits are –

  • Increases task precision

First of all, you should check which parameters define your SOTA Model. These parameters could be the recall or the precision, or the area under the curve (AUC). It could be any metric you choose. After that, you could determine the value of the SOTA for each of the chosen metrics. If these metrics get a high score (about 90%-95%) in performance accuracy, then it is labelled as a SOTA. Now it is pretty obvious that these models score high on accuracy, so the AI task will be as close to what the users need to do.  

  • Increases reliability

Since the precision of the SOTA models is high, as mentioned above, the reliability of the AI task also increases. If it is a machine learning task or deep neural network task, then be assured that the results are pretty much what they are supposed to be. They can be trusted and not be considered a random test of sorts. But how do you know that the SOTA is trustworthy?

So, here’s a suggestion. While you are building the SOTA test, it would be better if you ran noise experiments on the SOTA model. It will help you in measuring the standard deviation in the many identical tests runs that you are subjecting the model to. You can use this measurable deviation as a sort of shift or tolerance, and then you can compare the original SOTA result and the reproduced result. Testing the results will help you in verifying the features that are required in the algorithm in the future.

  • Ensures reproducibility

If you want your AI product to be agile and lean, then you will be able to ship the minimal viable product (MVP or a minimal version of your envisioned product) quickly to all your customers. You can then proceed to get user feedback and improve iteratively. Therefore, reproducibility in your SOTA model can be considered to be a good practice. This will help you in making compromises in your algorithm. You can also ship your algorithm quickly. And yes, about the customer feedback you have collected, you can use it as a guide for all your efforts in future product improvements.

  • Reduces generation time

Since the SOTA model helps you in reproducibility of the algorithm or the product, it also helps you in saving time when you put the entire process on the conveyor belt. That means you can make a saleable product from a prototype in less time than when you made the same product from scratch. All you need is to reproduce the algorithm on the parameters on which it needs to be tested are already in possession, so yes, you save a lot of time in the generation of the product.

When should you run a SOTA test?

You should run SOTA tests as frequently as possible. Frequent SOTA tests are a rule of thumb in AI. But it is advisable to run them once a week. You should also run the SOTA tests when you are incorporating important changes. It is advisable to run the SOTA tests should be run on a cloud virtual machine using a good pipeline like Jenkins.

Where can the SOTA models be used?

SOTA models are used in various artificial intelligence activities –

(a) Object detection by deep neural networks

(b) Single shot multi box detectors

(c) Self-adaptive tasks like choosing variable patterns

This list is not exhaustive as the possibility of using SOTA encompasses many branches of AI. Be on the lookout for future blogs to know more about SOTA and its applications in every subset of AI.

To sum up, SOTA models have played a crucial role in advancing AI and ML technologies. It has introduced structural efficiency that has boosted performance. Now, developers run various SOTA tests using the virtual GPUs, which further streamlines the process and reduce the upfront infrastructure costs, and E2E Networks is making it possible with cloud GPUs.

Reference Links

https://towardsdatascience.com/software-design-patterns-and-principles-for-a-i-1-sota-tests-3dd265c6bf97

https://deci.ai/blog/sota-dnns-overview/

https://paperswithcode.com/sota

This is a decorative image for : Should you migrate to E2E Cloud from Digital Ocean?
August 5, 2022

Should you migrate to E2E Cloud from Digital Ocean?

There comes a time in professional business life when they want to migrate all their data, resources, applications, workloads, etc., to the cloud for security reasons. It is a process of transferring data from on-premises to the cloud. Everyone prefers to use the cloud these days, but cloud migration can be an overwhelming process. Business wants to go with a service with minimal downtime and a hassle-free experience. So if you are using Digital Ocean for a while and now prefer to switch to another service, then this one question must have popped into your mind: should you migrate to the E2E cloud from the digital ocean? Here in this blog, we are going to answer the same. But first thing first, let's understand the benefits of migrating to the cloud.

The Top Benefit of Migrating to the Cloud

Businesses prefer to rely on cloud platforms due to various reasons, some of which are listed below:

1. Security

The first benefit of using a cloud platform is the high level of security compared to other network systems. The shared responsibility model is used in the cloud system, which is why this model is more successful than the traditional network system. All the data and resources of the business are stored centrally, which makes the cloud network convenient.

2. Scalability

The second benefit of using cloud platforms is scalability which means businesses can increase and decrease their requirement anytime based on the need and performance of the company. Firms and organisations have the flexibility to alter their infrastructure needs and workloads based on the current condition.

3. Integration

Another benefit of switching to a cloud platform is seamless integration. Businesses can connect multiple systems altogether without any difficulty. Not only does this increase the efficiency of the company, but it also saves money. Cloud services are updated and improved regularly, so the chances of decreased efficiency are less.

4. Cost

Lastly, one of the significant benefits of using cloud networks is cost. It reduced operational costs. Business here only pays according to usage, saving a lot of money.

Should You Migrate to the E2E Cloud From Digital Ocean?

Yes, that is possible, and with the recent hike in Digital Ocean's price, the only convenient option for organizations is to migrate to the best affordable solution. And when we talk about affordability, the E2E cloud seems to be the best in the market. The answer has a high-quality infrastructure. Around ten thousand clients are relying on E2E cloud solutions. The platform is built to fulfil the need of every kind of business. The solution is designed to execute real-world use cases such as NLP, health tech and consumer tech.

The thing which is loved by businesses is the quick deployment process, and the E2E cloud understands that very well. That's why companies will get the one-click deployment at their fingertip. And most importantly, the pricing model business will bring to the E2E cloud is unbeatable.

The process of cloud migration can be excruciating. Newbies and newcomers can't do it without proper assistance. That's why the E2E cloud is readily available at their customer service with the required help.

Sign up to E2E Cloud Now

With all that in mind, if you are looking for the most convenient solution, then give the E2E cloud a try. Not only will we help you save money, but you will get the cloud platform with high reliability too. Reach out to us to get a consultation on the migration process from Digital Ocean to our forum.

References

https://touchstonesecurity.com/cloud-migration-benefits/

https://www.vultr.com/news/Should-you-switch-from-DigitalOcean-to-Vultr/

This is a decorative image for- How do data scientists use PyTorch?
August 4, 2022

How do data scientists use PyTorch?

PyTorch was introduced for the first time in 2016 and it is a deep learning open-source framework. It has become very popular among developers due to its ease of usage and efficiency. PyTorch is getting huge critical acclaim because of its compatibility with a high-level programming language Python which is also favored by data scientists and machine learning developers.

About PyTorch

Deep learning models are a type of machine learning model that have multiple applications and usage which include language processing, image recognition, and more. PyTorch is an elegant framework that can help in the construction of deep learning models. This framework has been written using Python and the best part about PyTorch is that it is extremely easy to learn and implement for machine learning developers.

Furthermore, PyTorch is unique in its support of GPUs. Other exclusive features of PyTorch include auto-differentiation, reverse-mode, computational graph, etc. This is also why PyTorch is a popular choice among developers for prototyping and fast experimentation.

Why is PyTorch a popular choice among developers and data scientists?

PyTorch is the product of Meta’s Artificial Intelligence research lab and others. The framework has incorporated the Python programming language in the front end with a resilient and productive backend library from Torch which is also GPU accelerated. The entire framework concentrates on unreadable code, quick prototyping, and assisting multiple categories of deep learning models. 

Although PyTorch enables the friendly yet authoritative programming approach for data scientists and developers, simultaneously providing production graphs. The framework was released as open source in the year 2017 and because of its Python roots, it has become fairly popular among machine learning programmers.

Benefits of PyTorch for data scientists

Due to its innovative characteristics, PyTorch is extremely popular in deep learning. For example, PyTorch has implemented a chainer technology known as reverse-mode automatic differentiation. To put it more simply, the method is like a tape recorder that completes each and every operation, then computes the gradients, and finally iterates the entire process. 

Due to this particular feature, debugging in PyTorch is very simple and it can also adapt to specific applications such as dynamic neural networks.  PyTorch is also well accepted for prototyping because every repetition can provide different results.

Python developers extensively use PyTorch which has been developed using the Python language. The framework utilizes the define-by-run eager execution mode and authoritativeness of the language through which all the operations are executed. 

Although Python is fairly popular among developers and other programming languages, a recent survey by Datanami shows there has been a growing focus on machine learning, deep learning, and AI thus paving the way for industry-wide PyTorch implementation.

For existing Python developers and data scientists, PyTorch has become a good choice for its futuristic scope. Moreover, those who are comparatively new to deep learning can already come across an enlarging library of deep learning courses which are specifically based on PyTorch. Since its release, the API of this framework has remained consistent and that is why PyTorch is significantly easy to decipher for experienced Python programmers.

If we look at any particular strength of PyTorch then it is prototyping in smaller projects. It is also beneficial for academics and research communities because of its ease of usage and flexibility. Facebook’s AI research lab is also working tirelessly to ameliorate the productive application of PyTorch.

The latest releases of PyTorch have included multiple enhancements. Moreover, it has also added ONNX, or Open Neural Network Exchange which can help the developers comply with the deep learning models that will be productive for their projects or applications.

Features of PyTorch

Here is a list of important features of PyTorch:

 

  • PyTorch has an excellent and active community of developers that provides brilliant tutorials and documentation. You can visit their forum at PyTorch.org.
  • The entire framework has been developed using the popular programming language Python and the developers have also included Python libraries such as NumPy to conduct scientific computing. For the compilation of Python to C and to provide a better performance, SciPy and Cython have been used. 
  • PyTorch is very easy for data scientists and Python developers because it has similar syntax and utilization.
  • Major cloud platform supports PyTorch.
  • The scripting language of PyTorch is known as TorchScript and it is very easy to use as well as ductile when used in eager mode (eager mode is a specific mode of this framework where operations are executed instantly as they are derived from Python). You can also change to the graph mode if you require better optimization and more speed in C++ runtime settings.
  • PyTorch can effectively support parallel processing, GPU, distributed training, and CPU, which means any computational work can be allocated among various GPU and CPU cores. Furthermore, you can also conduct training on multiple machines using multiple GPUs.
  • Dynamic computational graphs are supported by PyTorch which enables the network behavior to be transitioned during the runtime. This flexible characteristic is a major feature that sets apart PyTorch from the existing deep learning frameworks (because the rest of them require neural networks to be delineated as a static object before runtime.)
  • PyTorch also has a storage of pre-trained models that can be replicated using a single code line.
  • PyTorch as a deep learning framework has both the eager mode (for experiments) and graph mode (for the execution of performance).
  • You can extend the core functionality of your applications using the brilliant APIs of PyTorch.
  • The libraries and tools of PyTorch range from reinforcement learning to computer vision.
  • The pure C++ frontend interface which the python developers are accustomed to is supported by PyTorch and you can also create high-performance C++ applications using the same.
  • In PyTorch, you will be easily able to construct a brand-new custom component as a subclass under the standard Python class.
  • You can easily import the libraries and parameters which can further be efficiently dispensed with the help of TensorBoard (which is an external toolkit.)

Practical use case of PyTorch for data scientists

Due to the PyTorch framework being convenient and flexible, it is being used in multiple projects and applications such as natural language processing, reinforcement learning, image classification, etc. Let us discuss them in brief:

Natural Language Processing (NLP)

If we look at software or virtual assistants, we will be able to understand how machine learning has made significant breakthroughs in understanding natural languages. 

Most of these models utilize a flat sequence of characters or words in the form of recurrent neural networks or RNN to process the sequences. Yet, a lot of linguistics think that language can be comprehended most efficiently if we use a stratified tree of phrases.

That is why a lot of research has been done on the deep learning models which are termed as recursive neural networks that undertake this approach recommended by linguistics. Although these models do have a complex nature and are hard to implement, PyTorch smoothens these difficult natural language processing models to make them much easier and more efficient. Right now, Salesforce is utilizing PyTorch for multi-task learning and NLP.

Computer vision

You can utilize computer neural networks to reinforce the development of image classification, object detection, and generative application. The framework also helps the programmers to process images and videos through which they will be able to construct a detailed and unambiguous computer vision model.

Reinforcement learning

You can easily control the motion of robots, create business development plans and reinforce robotic processes with the help of PyTorch.

How data scientists can work with reinforcement learning with the help of PyTorch

For data scientists, there are multiple use cases of PyTorch in the deep learning field. Moreover, you can experience better results with the implementation of PyTorch in multiple projects regarding style transfer, image classification for identifying fake goods, etc. 

Currently, tech giants are also using PyTorch for natural language processing. If we carefully look at the progress and implementation of PyTorch in the field of deep learning and artificial intelligence, learning this framework as one of your technical abilities can open up lots of future opportunities for you.

Reference links:

https://medium.com/geekculture/how-pytorch-helps-data-scientists-in-reinforcement-learning-a8843e441c1

https://towardsdatascience.com/minimal-pytorch-subset-for-deep-learning-for-data-scientists-8ccbd1ccba6b

Build on the most powerful infrastructure cloud

A vector illustration of a tech city using latest cloud technologies & infrastructure