MissingLink.ai has launched as a comprehensive deep learning lifecycle management platform allowing data scientists and engineers to significantly shorten the time it takes to train and deliver effective business outcomes.
With cameras in homes, retail spaces and in the pocket of almost every person on the planet, the volume of HD image and video data is creating a rapidly growing unstructured dataset. IDC forecasts that the amount of global data subject to data analysis will grow by a factor of 50 to 5.2 zettabytes in 2025.
At the same time, machine learning promises to save and improve lives in profound ways: better medical imaging detection for cancer and heart disease; safer roads through autonomous vehicles; more secure public spaces via crowd and body cams; easier grab and go shopping.
While there has been much chatter about the promise of machine learning to unlock the value of this ever-growing collection of data, the truth is that today, assessing data is too time-consuming and expensive. Engineers are spending too much time managing the sheer volume of data, instead of actually learning from it and being empowered to make changes.
These data scientists, who are in high demand, are in effect spending their time doing what may be thought of as DevOps, but are more tuned to the task of deep learning. Deep Learning Operations˜DeepOpsa new combination of cultural philosophies, practices and tools for AI developers, will enable organizations to create and automate their machine learning at a faster pace. Similar to DevOps, DeepOps underlines the importance of integration and delivery practices dedicated to Deep Learning workflows.
Stop Wasting Time on Menial Tasks, Focus on Solving Problems
MissingLink was born out of a desire to flip this problem, to allow teams of data scientists and engineers to spend their time solving world-changing problems instead of doing menial tasks.
Were at an incredible tipping point with all the data we need to solve really important problems, like saving lives through cancer detection and providing safer, smarter driving on the streets. But wading through all that data to find the meaning from it is tough and requires too much manpower, said Yosi Taguri, co-founder of MissingLink.ai. MissingLink allows every engineer to build complex AI machines in a way that wasnt possible before. Were taking away a lot of the grunt work, so they can focus on the bigger picture issues.
How MissingLink Helps Data Engineers
Developing and running a deep learning workflow is resource-intensive and time-consuming. It requires managing multiple datasets and versions of models, including data, experiments, compute resources and code.
With MissingLink, data teams can:
- Get started with 3 lines of code: Setting up an experiment requires tedious work including log parsing, copying data, managing machines, running experiments manually and logging the analysis. With MissingLink, just three lines of code lets you effortlessly integrate code, data and existing infrastructure.
- Manage data like source control: While AI is one of the most cutting-edge fields in computer science, the industry is still using the same old tools like file systems. MissingLink offers a version-aware data store, eliminating the need to copy files and only syncing changes to data, resulting in reduced load time and easy data exploration.
- Get to results faster: Environment and resources required for running experiments at scale only need to be set up once. Experiments can be provisioned to run in advance in an automated fashion with the ability to automatically scale resources up/down as needed. MissingLink allows for easy monitoring with real-time tracking of experiments via visual dashboards, providing the ability to make decisions in real time.
- Reproduce experiments easily: The ability to reproduce experiments is critical for understanding problems and optimizing solutions. MissingLink automatically tracks all of your data, experiments and code, allowing you to easily reproduce any experiment at any time.
- Increase productivity: As teams run deep learning experiments, they have to track and manage countless elements, versions and data points. This is repetitive and tiresome work that is both extremely frustrating to individuals, as well as a massive loss of time and money for an organization. With MissingLink, some customers have been able to run ten times more experiments in the same time, even running dozens of experiments simultaneously.
- Manage hybrid resources with a single command: MissingLink allows teams to manage both local and public cloud resources as a single environment, allowing you to grow and shrink your compute resources elastically as needed.
- Keep data and code with you: There is no need to upload your experiment data and code, keeping it secure and private, and helps ensure industry compliance.
- Easily handle large–scale datasets: Computer vision deep learning projects bring additional challenges. The size of very large image and video files makes them expensive to store and running experiments on them can take a very long time. MissingLink provides data management at scale, allowing companies to keep their data onsite and adjust experiment usage as your needs change.
Data engineers at companies including Aidoc, Nanit and Way2VAT have been in production with MissingLink.ai for the past 12-18 months and are already experiencing faster results, higher quality outputs, and easier management of deep learning work.
MissingLink.ai is exactly what we needed for deep learning at scale, said Idan Bassuk, head of AI at Aidoc, which develops an FDA-approved AI-based software for medical imaging. We hire top data science talent to focus on high-quality algorithms and engineering. Running dozens of experiments per day requires hours of DevOps work, maintenance and idle time. MissingLink’s solution enables 10 minutes of preparation and a click of a button.
Every month we run hundreds of experiments on millions of image data points, said Tor Ivry, founder and CTO of Nanit, maker of the popular smart baby monitor that uses deep learning and computer vision. MissingLink.ai is one of the only solutions for data management that can handle this scale, saving us a lot of time and making it super-easy to manage data versions.
Availability and Pricing
MissingLinks deep learning platform is available now and starts with a free version that you can scale as you grow.
MissingLink.ai is a powerful deep learning platform that helps data engineers streamline and automate the entire deep learning cycle: data, code, experiments and resources. It eliminates the grunt work and significantly shortens the time it takes to train and deliver effective models. MissingLink.ai is used by data engineers at companies including Aidoc, Nanit and Way2VAT. MissingLink is a part of Samsung NEXT. Learn more at MissingLink.ai and Samsung NEXT.