This year, IBM Think 2018 was hosted in Las Vegas from March 20 to 22. It was one of the most anticipated IBM events in 2018, with over 40,000 developers as well as technology and business leaders in attendance. Considered IBM’s flagship conference, Think 2018 combined previous conferences such as IBM InterConnect and World of Watson.
IBM Think 2018: Key Takeaways
- IBM Watson Studio announced – A platform where data professionals in different roles can come together and build end-to-end Artificial Intelligence workflows
- Integration of IBM Watson with Apple’s Core ML, for incorporating custom machine learning models into iOS apps
- IBM Blockchain platform announced, for Blockchain developers to build enterprise-grade decentralized applications
- Deep Learning as a Service announced as a part of the Watson Studio, allowing you to train deep learning models more efficiently
- Fabric for Deep Learning open-sourced, so that you can use the open source deep learning framework to train your models and then integrate them with the Watson Studio
- Neural Network Modeler announced for Watson Studio, a GUI tool to design neural networks efficiently, without a lot of manual coding
- IBM Watson Assistant announced, an AI-powered digital assistant, for automotive vehicles and hospitality
Here are some of the announcements and key takeaways which have excited us, as well as the developers all around the world!
IBM Watson Studio announced
One of the biggest announcements of the event was the IBM Watson Studio – a premier tool that brings together data scientists, developers and data engineers to collaborate, build and deploy end-to-end data workflows. Right from accessing your data source to deploying accurate and high performance models, this platform does it all. It is just what enterprises need today to leverage Artificial Intelligence in order to accelerate research, and get intuitive insights from their data.
IBM Watson Studio’s Lead Product Manager, Armand Ruiz, gives a sneak-peek into what we can expect from Watson Studio.
Collaboration with Apple Core ML
IBM took their relationship with Apple to another level by announcing their collaboration to develop smarter iOS applications. IBM Watson’s Visual Recognition Service can be used to train custom Core ML machine learning models, which can be directly used by iOS apps.
The latest announcement at IBM Think 2018 comes as no surprise to us, considering IBM had released new developer tools for enterprise development using the Swift language.
IBM Watson Assistant announced
IBM Think 2018 also announced the evolution of Watson Conversation to Watson Assistant, introducing new features and capabilities to deliver a more engaging and personalized customer experience. With this, IBM plans to take the concept of AI assistants for businesses on to a new level.
Currently in the beta program, there are 2 domain-specific solutions available for use on top of Watson Assistant – namely Watson Assistant for Automotive and Watson Assistant for Hospitality.
IBM Blockchain Platform
Per Juniper Research, more than half of the world’s big corporations are considering adoption of or are already in the process of adopting Blockchain technology. This presents a serious opportunity for a developer centric platform that can be used to build custom decentralized networks. IBM, unsurprisingly, has identified this opportunity and come up with a Blockchain development platform of their own – the IBM Blockchain Platform.
Recently launched as a beta, this platform offers a pay-as-you-use option for Blockchain developers to develop their own enterprise-grade Blockchain solutions without any hassle.
Deep Learning as a Service
Training a deep learning model is quite tricky, as it requires you to design the right kind of neural networks along with having the right hyperparameters. This is a significant pain point for the data scientists and machine learning engineers.
To tackle this problem, IBM announced the release of Deep Learning as a Service as part of the Watson Studio. It includes the Neural Network Modeler (explained in detail below) to simplify the process of designing and training neural networks. Alternatively, using this service, you can leverage popular deep learning libraries and frameworks such as PyTorch, Tensorflow, Caffe, Keras to train your neural networks manually.
In the process, IBM also open sourced the core functionalities of Deep Learning as a Service as a separate project – namely Fabric for Deep Learning. This allows models to be trained using different open source frameworks on Kubernetes containers, and also make use of the GPUs’ processing power. These models can then eventually be integrated to the Watson Studio.
Accelerating deep learning with the Neural Network Modeler
In a bid to reduce the complexities and the manual work that go into designing and training neural networks, IBM introduced a beta release of the Neural Network Modeler within the Watson Studio. This new feature allows you to design and model standardized neural network models without going into a lot of technical details, thanks to its intuitive GUI.
With this announcement, IBM aims to accelerate the overall process of deep learning, so that the data scientists and machine learning developers can focus on the thinking more than operational side of things.
At Think 2018, we also saw the IBM Research team present their annual ‘5 in 5’ predictions. This session highlighted the 5 key innovations that are currently in research, and are expected to change our lives in the near future.
With these announcements, it’s quite clear that IBM are well in sync with the two hottest trends in the tech space today – namely Artificial Intelligence and Blockchain. They seem to be taking every possible step to ensure they’re right up there as the preferred choice of tool for data scientists and machine learning developers. We only expect the aforementioned services to get better and have more mainstream adoption with time, as most of these services are currently in the beta stage. Not just that, there’s scope for more improvements and addition of newer functionalities as they develop these platforms.
What did you think of these announcements by IBM? Do let us know!