Weekly Digest #16: Google to Help with Speech Recognition
July 28th, 2016
Last week Google has introduced a few new interesting products in the field of artificial intelligence and speech recognition and IT market leaders have created a secure protocol for the Internet of Things. Meanwhile, IBM & Macy’s are testing artificial intelligence usage in customer support.
Robot as a Financial Advisor
Robots can now not simply help with daily activities, but also give advice on investments. Next year, a robot will consult regarding investments in one of the world’s largest banks - Wells Fargo & Co. This robot is based on a series of algorithms that help give recommendations to clients. The tips and advice will be given, taking into account the available information about the customer - for example, his / her willingness to take risks (and do high-risk investments).
Almost-Siri for Customers
Well-known retailer Macy’s and computers’ giant IBM announced the launch of a pilot project called “Macy’s On Call.” It is a mobile tool that allows customers to interact with the artificial intelligence platform through their mobile devices. In a way, “Macy’s On Call” is something like Siri for customers of the retail stores. The interaction is done via Satisfi - intelligent customer engagement platform that uses geo-location for work. The solution will be deployed in 10 test locations across the United States.
According to the company, the service will allow customers to ask questions about products, services, or facilities, and receive personalized answers. For example, the client might ask about the location for the women’s shoe store and specify a particular brand or manufacturer. In response, Macy’s On Call will tell where to get one’s favorite pair of shoes or sneakers, choosing the stores closest to the client’s location.
The pilot project will allow customers to communicate with Macy’s On Call not in English only, but also in Spanish. This would provide a wider coverage for customers and make the interaction more convenient for those, for whom English is not a native language.
IBM provided its IBM Watson technology for the project. Watson is a platform that uses the tools for natural language recognition and processing as well as machine learning tools for analysis and further use of large volumes of unstructured data.
Google helps to recognize speech
Google has made available a new interface called Cloud Natural Language API for developers. This instrument can recognize human speech, parse it, and help to understand its meaning by breaking it into pieces and finding information about people, places, events, etc. For example, you can find a review about your product in a social network or find all the people in Harry Potter books with a single API call. New tool complements the family of interfaces of machine learning platforms Vision API and Translate API.
Google created its Cloud Natural Language API based on a similar design they used to create Parsey McParseface. The API can work with texts in English, Spanish, and Japanese. All these tools will help developers to create chatbots and intelligent digital assistants. At the moment, one can analyze up to 5,000 articles for free per month.
Another product is announced yet not fully revealed yet - Google Speech API set, that allows applications to convert audio into text in more than 80 different languages. This API can be integrated into the Internet of Things devices.
OTrP - Internet of Things Security Protocol
Internet of Things devices and platforms are becoming more and more popular and with their popularity, the issue of security & safety is growing in its importance as well. Major IT companies have turned their attention to this challenge. ARM & Symantec, as well as several other companies, have developed a new security protocol called Open Trust Protocol (OTrP). Its mission - to provide a secure architecture and code management for the connected devices’ protection.
OTrP can be implemented in systems based on public key infrastructures. Thanks to this, application developers and device manufacturers are now about to use their own keys for app & devices authentication and management. OTrP can be easily integrated into existing Trusted Execution Environments or platforms that support RSA-cryptography.
Interested in taking a look? You can download the OTrP prototype from the website of Internet Engineering Task Force.
Artificial Intelligence Takes Care of Energy Conservation
Google’s artificial intelligence reduces the costs of cooling the company’s data centers. DeepMind, machine learning system, is used for the energy consumption analysis and it has created algorithms that lowered the resource usage as a result. Cooling systems now use 40% less energy and overall electricity, consumed by the entire building, fell by 15%. DeepMind network works in real time - it receives data about temperature, power consumption, pumps speed, and adjusts all parameters based on the received data.
Google plans to use these algorithms and technology in its other divisions. Besides, Google promises to provide the information about these instruments’ operation to other data center owners so they can use them in their work as well, working towards a better future for everyone.
Neural Networks can help you buy “Same Dress, but Different”
Russian-based Yandex has presented an application that allows purchasing clothing similar to the one shown in a picture. The application called “Take off your clothes” uses neural networks for image recognition and identification of clothing items. After that, the system searches for similar clothes in stores. The user can simply take a picture (or choose from a gallery) of a clothing item he/she likes and the app will find counterparts. The app can also choose clothes based on color, size, or budget.
App authors say that two neural networks are used for the image processing. One is for recognizing images in general, while another one is used for identification of the clothing item itself - for example, that image contains a picture of jeans of dark blue color.
Augmented reality comes to the corporate sector
Analysts forecast the growth of augmented reality technology enterprise market. One of the main development directions for this segment will be the provision of remote assistance. It will be implemented using special devices development - for example, smart glasses for the corporate sector. These glasses can be used, for example, in healthcare, manufacturing, or even in public services field. The report made by ABI Research notes that smart glasses supply will reach up to 27 million devices by 2021. However, these glasses will have to provide high-quality display abilities, have long battery life on a single charge, and be user-friendly. The forecast, nonetheless, doesn’t seem that far-fetched considering the AR-ecosystem of applications and content for them is growing every day.