Commercial AI vs Open Source: What You Should Choose?
regens.com

Commercial AI vs Open Source: What You Should Choose?

Disclaimer:

This article is cut by two parts, first part is on Acknowledge (Summarized by ML Buyers Guide 2018 by ActiveState) and second part is Commercial AI vs Open Source (The actual considerations that answer questions on the title)


Acknowledge

"Artificial intelligence will, on average, boost rates of profitability by 38% and provide an economic boost of $14 trillion in additional gross value by 2035" - Accenture


Next upcoming years will be defined by the rise of Artificial Intelligence (AI). AI is driving the creation of systems that cannot only to learn and adapt, but may at some point even act autonomously. Even though AI is so sophisticated, current technology in AI is primarily being used to augment and improve human activity via Machine Learning (ML); where ML is focused on giving machines access to data and letting them learn for themselves.

Based on data collected in the enterprise, ML is being used today to enhance decision making, disrupt traditional business models, and redefine the customer experience.

ML Driven Company

All companies today are data companies, whether they realize it or not. As the new oil, data has become the core of many organizations digitization efforts. Enterprise are focusing on what data to collect along with how to collect it. They are hoping to follow in the footsteps of companies like Google, Facebook, Alibaba, Tencent, Amazon, Baidu, Uber and Airbnb. Because all of these have gained market advantage by being able to better analyze, monetize and gain new insights from their business data by using Machine Learning.  


If you find yourself in the planning stages for ML but not ready to implement, you’re not alone. Only about 12% of our customers have either implemented data science and ML, or are implementing it right now . - Gartner


Data scientists, on the other hand, tend to be not only math and statistics focused, but also highly skilled at finding business values in data. Unfortunately, demand currently outstrips supply for these individuals, especially for those that also have either programming and/or specific domain skills. If you’re lucky enough to hire people that has this capacity, you can centralize all your data science and ML efforts in order to go deep and get the biggest profitability on your investment. If not, it may make more sense to go wide, by nurturing the skills you already have across your organization. In fact, by growing data science skills in every department, you can apply machine learning closer to the domain problems. And therefore scale ML across the organization. While in-house development may sound like a pipe dream, there have been recent advances in a number of great tools and platforms that make ML more accessible than ever to a wide range of IT.


The global cognitive computing market is expected to reach $12.5 billion in 2019, up from 2.5 billion in 2014, at a CAGR of 38%. - McKinsey


A comprehensive ML solution (like we build in dmand.ai) provides features and functionality that encompass: (1) data preparation, like data pooling, cleansing, labeling, encryption, and anonymization etc. (what we call it Datanest engine) . (2) Data science (e.g., algorithms and model building), and (3) operationalization (i.e., incorporating ML models into software applications, using containerization and orchestration approach).

The data science effort is very compute-intensive. By comparison, the data preparation work is labor-intensive. As a result, a significant proportion of your organization’s work will be focused on preparing the data: organizing it into appropriate domains, cleansing it of bad data, finding subject matter experts to label the data correctly, and creating the data pipelines that will feed your model.

In a nutshell, ML is all about math, specifically statistics and algorithms, two areas that have been addressed by software solutions for decades. What’s new is the price tag and seamless integration.

Commercial AI vs Open Source

Previously these kinds of solutions used to be prohibitively expensive to purchase, deploy and maintain, but today organizations can choose from either a growing number of affordable commercial solutions or freely downloadable open source options. 

The advantages of open-source is the community, at least it cover in 4 areas:

  1. Code and documentation availability, website like Github and original page of the open-source repository give code and documentation available freely,
  2. Open source does not require a license fee, and therefore attracts a significant attention from community
  3. Large open source communities provide routine bug fixes and language simplifications, as well as answers to common questions in implementation, you can see StackOverflow to know more details about which community is active and which community isn't
  4. A large community also makes it easier to recruit technical resources, sometimes freshers who are experts in using the latest technology.

But, open-source is still open-source, no one can guarantee it matched with what you need. By using Commercial AI (like dmand.ai), you can get some additional benefits:

  1. SLA-backed support and maintenance, Open source software (OSS) developed by collaborative work done by volunteers through around the world with different management styles. Open source code is updated and modified all the time from the first release. Therefore, there is a need to measure the quality and specifically the maintainability of such code. 
  2. Time savings: by pre-defined and pre-integrated functionality you can save time over cobbling and integrated together a set of open source modules from various authors
  3. More cost efficient for domain specific but generic implementation (like dmand.ai in retail, warehouse management, and salesman scheduling).
  4. Give data literacy boost, detail can be seen in this Post .

Basically, the choice is yours, commercial solution is always give benefit. However, the choice is based on your needs.

vignesh .

Microsoft Practice | Photographer

4 年
Anantha Padmanabhan S S

Business Analytics | Product Management | HR | L&D | NeuroLeadership

4 年

Great piece NABIH IBRAHIM BAWAZIR

回复
Kunal Sinha

Quality Assurance Specialist at PTC

4 年

What is your opinion on AWS Sagemaker

要查看或添加评论,请登录

NABIH IBRAHIM BAWAZIR的更多文章

  • My FAQ and Answers

    My FAQ and Answers

    Disclaimer : This is 6th April 2020 condition. Sorry for being impolite, I just try to be honest.

    3 条评论
  • A-Z in Data Science

    A-Z in Data Science

    This is just dictionary, i will update if I found good article to read A/B Testing Data Science Interview Machine…

    22 条评论

社区洞察

其他会员也浏览了