Cloud TPUs available in beta, Microsoft’s plans for blockchain, Oracle’s PaaS, Scanpy, and more in today’s top stories around machine learning, blockchain, and data science news.
Google’s Cloud TPUs, the family of Google-designed hardware accelerators, is now available in beta. These custom chips are optimized to speed up and scale up specific ML workloads programmed with TensorFlow. The company first announced Cloud TPUs at its I/O developer conference on May 17-19, 2017, for a limited number of developers and researchers.
Each Cloud TPU features four custom ASICs. It packs up to 180 teraflops of floating-point performance and 64 GB of high-bandwidth memory onto a single board.
Developers who already use TensorFlow don’t have to make any major changes to their code to use this service. Usage will be billed at $6.50 per Cloud TPU per hour. Using a single Cloud TPU, developers can train ResNet-50 to the expected accuracy on the ImageNet benchmark challenge in less than a day, for well under $200! Once the TPU pods (array of cloud TPUs connected together via an ultra-fast, dedicated network to form multi-petaflop ML supercomputers) are available, ResNet-50 and Transformer training times will drop from almost a day to less than 30 minutes.
In a recent blog post, Microsoft revealed plans to use blockchain technology in the form of decentralized identity systems to solve online management issues with personal data and identity management. A decentralized identity system is not controlled by any single, centralized institution. It removes the possibility of censorship and gives an individual full control over their identity and reputation. The building of the platform takes inspiration from Microsoft’s commitment to the ID2020 alliance. Initially, Microsoft will support blockchain-based decentralized IDs (DIDs) through the Microsoft Authenticator app. Microsoft plans to work with DID method implementations, which follow a specific standard outlined by a W3C working group. According to Ankur Patel, PM, Microsoft’s identity division, “Using our technology individuals will get a secure encrypted digital hub where they can store their identity data and easily control access to it.”
Oracle lays out a broader vision for Oracle Cloud Platform with a range of autonomous service capabilities. Oracle PaaS (Platform as a service) capabilities support the needs of the entire organization, including developers, enterprise architects, data scientists, IT operations, and business users.
- Autonomous PaaS services include advanced capabilities such as auto code generation, self-defining data flows, automated data discovery and preparation.
- They also have voice-enabled integration links, machine learning-based continuous data analysis, and self-learning bots that understand user intent and continually refine that understanding.
- They will speed IT deployments, by letting developers jump right to creating new functionality rather than having to spend time on the routine tasks.
- PaaS promises to lower IT costs and improve security because they require less human management and eliminate human error.
Scientists from the Helmholtz Zentrum München have developed Scanpy, a program that is able to help manage enormous datasets. Scanpy was made with the purpose of analyzing the gene-expression data of a large number of individual cells. It allows comprehensive analysis of large gene-expression datasets with a broad range of machine-learning and statistical methods. Scanpy is based on the Python language and uses graph-like coordinate system. Instead of characterizing a single cell by the expression value for thousands of genes, the system simply characterizes cells by identifying their closest neighbors — very much like the connections in social networks. In fact, to identify cell types, Scanpy uses the same algorithms as Facebook does for identifying communities. To read more, visit the official documentation.
5. Accelirate has announced its partnership with Chirrp.ai to strengthen its enterprise-class chatbot solutions capability
Accelirate has partnered with Chirrp.ai, an AI-powered communication channel provider, to strengthen their chatbot solutions. With this partnership, the enterprise-grade chatbots will be able to handle low-, medium- and high-complexity use cases. A high-complexity use case is where many clients use chatbots as an NLP/NLU-powered application-delivery mechanism which can handle complex user queries as well as application rules and workflows right from within the chatbot interface. Initially, the chatbots will be configured and set up to understand structured as well as unstructured customer queries and provide them with appropriate answers without involving a human. The chatbot will gather the relevant customer information, query the backend systems (which can be accomplished by using RPA robots) and present the information to the customer interactively. All without human intervention!