The First ML Value Chain Landscape
TheSequence recently released the first ever ML Chain Landscape shaped by data scientists, a new landscape that would be able to address the entire ML value chain.
Source: TheSequence
TheSequence is an ML community that has worked with DeepMind, OpenAI, Google Brain, and many more. The world of ML and AI is growing so quickly, to keep up with all of the changes is difficult for professionals and aspiring professionals. The goal at TheSequence is to make people who are interested in the sector smarter about AI without having to read white papers and spend hours on end researching.
They also offer other days of knowledge, if you would like to know more click on this link.
The First Ever Machine Learning Chain Landscape
On October 12th, 2022, TheSequence released the first-ever ML Chain Landscape shaped by data scientists. This sparked from TheSequence's wanting to create a new landscape that would be able to address the entire ML value chain, a project that was comprehensive but very valuable.
The project was made up of two sections:
- Preparing and using existing research
- Involving community members to evaluate and reshape the value chain
The 6 stages of a typical ML process include:
- Data Collection
- Data Processing
- Data Labeling
- ML Model Training and Evaluation
- ML Model Deployment
- Model Monitoring
TheSequence co-founder, Ksenia Se, states:
“We all thought the ML field had its problems, but we didn’t quite realize how disjointed it was. It seems that many ML practitioners either struggle or don’t reach their goals at all. And not because of poor code or even noisy data. But due to software incompatibility and poor strategizing within the chain that usually starts to assume form right from the get-go.”
Looking at the image below, the ML Value Chain Landscape produced by TheSequence tells us that unfortunately there is not one market solution that covers all six stages of the chain.
Vendors that are covering at least 5 stages include H2O, Dataiku, Clear ML, Vertex AI, Scale AI, Toloka AI, and Abacus.AI.
Click to enlarge
What the landscape revealed was that:
Around 50% of data processing and model monitoring is a struggle
This normally starts in the early stages of production and the challenges continue throughout till post-deployment. Important aspects of the success of an ML project also lie within the collection of data and the effectiveness of data labeling.
Poor optimization in the development phase
One of the root challenges is poor optimization in the development phase which highly reflects in the later stages. Due to this, the monitoring of a model is a stage in the value chain landscape that is highly avoided.
Data processing challenges
Data processing is never a simple task and is a very fragmented field, and it comes with its challenges. Many are eager to find a process that is easy to use, configure and scale; however, this has not been found yet.
Free-flowing better different stages
With anything you do, it’s nice when one thing is efficient to be used for the next stage and then the next stage. However, if there is no interaction or compatibility between the different stages - it results in a low ML project completion rate. There are a lot of software tools out there that are unfortunately incompatible and the more technical solutions get, the harder it is for professionals from different backgrounds to collaborate effectively.
Being able to cover the entire value chain
As seen in the image above, none of the vendors were able to cover the entire value chain, though some were very close. However, there is currently no flexible universal environment available that meets the demands of the spectrum of ML at every single stage.
Ksenia says:
“I'd like to add that since the publication of this landscape, we saw a lot of interest in it from our community. The ML industry is developing extremely fast, it's hard to keep up with all the startups that work on different parts of the ML Value chain. We plan to update this landscape regularly.”
Want to Know More?
If you want to stay up-to-date with the movements in machine learning, artificial intelligence, and the data science world - be sure to subscribe to TheSequence by clicking on this link.
Nisha Arya is a Data Scientist and Freelance Technical Writer. She is particularly interested in providing Data Science career advice or tutorials and theory based knowledge around Data Science. She also wishes to explore the different ways Artificial Intelligence is/can benefit the longevity of human life. A keen learner, seeking to broaden her tech knowledge and writing skills, whilst helping guide others.