ZeroML Platform — UX case study
‘0’ ML is a “ DATA SCIENTIST ”platform that can self-serve to create practical ML models capable of delivering promising results. In simple terms, it is the process of teaching a machine to deduct information in order to complete given tasks.
Building an AI or ML model that acts like a human requires large volumes of training data. For a model to make decisions and take action, it must be trained to understand specific information. Data annotation is the categorization and labeling of data for AI applications. Training data must be properly categorized and annotated for a specific use case. With high-quality, human-powered data annotation, companies can build and improve AI implementations. The result is an enhanced customer experience solution such as product recommendations, relevant search engine results, computer vision, speech recognition, chatbots, and more.
There are several primary types of data: text, audio, image, and video.
The project
To learn, the machine needs to be taught through training data.
For that project, the training data was written text that varied from online product reviews to survey responses to chatbot conversations and others.
The process of transforming the original data into training data involves manual annotations of the text. This is a long, painstaking and error-prone approach.
Text Annotation
The most commonly used data type is text — according to the 2020 State of AI and Machine Learning report, 70% of companies rely on text.
Sentiment Annotation
Sentiment analysis assesses attitudes, emotions, and opinions, making it important to have the right training data. To obtain that data, human annotators are often leveraged as they can evaluate sentiment and moderate content on all web platforms, including social media and eCommerce sites, with the ability to tag and report on keywords that are profane, sensitive, or neologistic, for example.
Audio Annotation
Audio annotation is the transcription and time-stamping of speech data, including the transcription of specific pronunciation and intonation, along with the identification of language, dialect, and speaker demographics. Every use case is different, and some require a very specific approach: for example, the tagging of aggressive speech indicators and non-speech sounds like glass breaking for use in security and emergency hotline technology applications.
Image Annotation
Image annotation is vital for a wide range of applications, including computer vision, robotic vision, facial recognition, and solutions that rely on machine learning to interpret images. To train these solutions, metadata must be assigned to the images in the form of identifiers, captions, or keywords.
From computer vision systems used by self-driving vehicles and machines that pick and sort produce, to healthcare applications that auto-identify medical conditions, many use cases require high volumes of annotated images. Image annotation increases precision and accuracy by effectively training these systems.
Video Annotation
Human-annotated data is the key to successful machine learning. Humans are simply better than computers at managing subjectivity, understanding intent, and coping with ambiguity. For example, when determining whether a search engine result is relevant, input from many people is needed for consensus. When training a computer vision or pattern recognition solution, humans are needed to identify and annotate specific data, such as outlining all the pixels containing trees or traffic signs in an image. Using this structured data, machines can learn to recognize these relationships in testing and production.
0ml.ai machine learning models are trained to leverage your data and can be easily integrated with any application using Rest APIs. We use state-of-the-art infrastructure, which scales intelligently according to the use cases.
The problem
How to annotate data fast and produce high-quality data annotation on a large scale?
There are existing tools that offer similar to 0ml.ai. But none of them provides the base of a scalable annotation process promptly, and those are not having advanced training detection models like as listed below.
A very talented group of engineers (my team) and I, the product designer, took the initiative to design an internal annotation platform, which will allow :
- image classification
- text classification
- sentiment analysis
- object detection
- chatbot
- face detection
- structure data modeling
- geographic information system
The final goal was to design a suite of products that will allow the any team or individuals to create practical ML models capable of delivering promising results as a fast and collaborative manner on the platform.
How do I fix this? — Brainstorming
My Proposal:
To fix the error and codeless effort of time-consuming data annotation methods, I came up with the idea of Design Patterns and UI Library. This feature would allow the user to learn easily and do it in their own way and it could look like a simple format, reliable, and minimal user interface that can be easy to navigate between one to another.
Also, What features give you the best UX for the desktop platform ?
- Providing companies to develop machine learning without coding and computer vision by our algorithms.
- 24x7 customer service assistance for every user to deploy the trained dta in the platform.
- Giving secure, speed, and scalable platform and in-built to handle large-scale data collection and annotation projects for the platform users. Also, to meet and exceed your quality standards for your training data.
Data labeling and task process in the platform
Data labeling required that we first craft the data text annotation tool.
We started our design system research and After several explorations, we studied two specific systems, which were abundant in useful components:
As the product designer, I laid the grounds for a special UI and Design system for the platform.
I began by following the Atomic Design Methodology, focusing on the necessities first: grid and layout, modular typography scale, colors, buttons, molecules, organisms, and pages. The aim was to craft templated flows, which can be used across the platform.
The design system is a work in progress and we continue to grow it in Adobe XD and Storybook.
The advantage of the tool is that multiple annotators can work on one dataset at the same time with different accounts. Also, It can call the API from the server then it performs automatic operations, which already written by engineers in the platform. annotators also can perform the custom model operation by training the data model for their large data set model like Vehicle to find the number plate for two and four-wheelers.
0ml.ai helps data analysts and project teams to work fast and in collaboration in order to train high-quality different types of data. The platform is stable and perfect. There is more to be desired but while building it we learned a lot. We managed to enable teams to train their data models easily and to find enjoyment in this painstaking process.
While doing so, we became experts in prioritizing goals, selecting the better design trade-offs, exposing and including other team members in our process, and most importantly, producing a tool that makes our users’ life a bit better or easier than other tools/platform.
