Image 29 October 2018
Image
 
Image

Research/ Public Policy

Reading China: Predicting Policy Change with Machine Learning

Researchers have developed a ML powered quantitative indicator of Chinese government’s policy priorities and has been termed as Policy Change Indicator (PCI). The developed algorithm read People’s Daily, official newspaper of Communist Party of China, from 1951 to Q3,2018. The paper is a medium for China’s party-state to present government directives, official standpoints and signaling future policies. The dataset on which the algorithm was trained, consisted of around 1,900,000+ articles published in the newspaper. Along with the article, the page on which they appeared were also recorded. ML techniques including word embedding, multilayer perceptrons and recurrent neural networks were deployed to detect changes in priority order of policy issues in the newspaper. Read More

Image
Image

Natural Language Processing

Tableau Launches NLP Powered Analytics Tool

Tableau has launched “Ask Data”, a natural language processing (NLP) tool which will enable customers to ask questions in form of a conversation. The algorithm is built keeping in mind customers’ intent of question rather than keywords used. The algorithm is capable of working with published data sources without any additional setup requirement. The new version will be able to anticipate needs of its users through the questions asked and provide relevant visualizations accordingly. Additionally, data modelling tools will automatically recognize relationships in data stored in Oracle and SQL server databases. Read More

Image
Image

Health Care

Blockchain and AI to Personalize Treatment Plans

Open Health Network, a health start-up, has launched “PatientSphere”. It collates data that contribute to patient’s health care record and then, derive personalized insights from them. The ML algorithm provides an adaptive treatment plans and exercise tips. Each tip is updated over time in accordance to patients’ progress rate. Compared to conventional post-operative programs suggesting textbook exercises of fixed duration and intensity, PatientSphere is capable of modifing treatment intensity as per requirement. Read More

Image
Image

Information Technology

Offline Speech Recognition with 97% Accuracy

Researchers at the University of Waterloo and startup DarwinAI have developed a strategy for speech recognition networks that is capable of running models on low-end smartphones. Accuracy level of these models is superior to its predecessors. Conventional deep learning approaches to voice recognition requires large powerful remote servers for processing. The new system is built on Amazon’s Alexa Machine Learning team work. Model performance was evaluated using Google Speech Commands dataset containing 65,000 one-second samples of 30 short words and background noise samples. Developed models performed better on accuracy, decreased latency and decreased memory requirement. Read More

Image
Image

Information Technology

World’s First AI on Blockchain Testnet

Cortex has launched its blockchain project that aims at providing ML models to customers to infer via smart contracts on Cortex blockchain technology. They have launched fully functional version of “Mining TestNet” and “AI Smart Contract TestNet”. The AI models have been developed on blockchain entirely. They are capable of automating contracts creation and automated decision making. It has been developed on open source technology. Cortex has developed 10 full TestNet nodes in 6 different countries to generate AI smart contract logs and inference results. Read More

Image
Image

Civil Services

Police Agencies Using AI to Spot Written Lies

Scientists from Cardiff University and Charles III University of Madrid have a tool named, “VeriPol” that focusses on spotting lies in written robbery reports. The team trained their ML model on over 1000 police robbery reports including those which were false. The algorithm identifies features in statements like adjectives, verbs, and punctuations marks, and then picks up on the patterns in false reports. In ints inference, false reports were found to be shorter, focused on stolen property rather than robbery, have fewer details about the attacker and lack witnesses. In a pilot study, VeriPol identified 64 false reports within a week’s time. Read More

Image
Image

Digital Marketing/Miscellaneous

AI System to Auto-Create a Logo Based on Colour Choices

Researchers from Maastricht University, Netherlands have developed an algorithm that can craft logos from 12 different colors. It has been termed as “LoGAN” for “Generating Logos with a Generative Adversarial Neural Network Conditioned on color”. The technology being used is “Generative adversarial networks” (GANs). Team trained the algorithm on dataset consisting of 486,777 small icons of 32 by 32 pixels. Dominant colour was extracted by an algorithm and was converted from RGB values to color words. Another neural network classifies the sample images before putting in in different categories. Read More

Image
Image

Miscellaneous

AI to Help in Designing Novel Perfumes

IBM and Symrise have together created first of its kind AI designed perfumes for sale. The system has been named as “Philyra” and is capable of learning perfume formulas, raw materials, historical success data and industry trends. It uses ML algorithms to go through several formulas and combinations of raw materials, helping identify patterns and perfect fragrance. The algorithm is also capable of developing new fragrance formulas. Philyra learns through a “distance model” to identify fragrances that are close in smell to existing fragrances. Larger the distance between a fragrance and its neighbours, more novel the perfume is predicted to be. Read More

Image
Image

“Quantum Computing, a revolution in technology, has shown great promise towards solving complex computing problems currently outside the capabilities of current computers. Though in its infancy, we at Decimal Point Analytics strongly believe that it is going to grow exponentially in the near future. It holds the potential not only to boost the AI revolution but also transform the way data is synthesized. The articles on Quantum computing will give an insight into the recent developments in this space.”

 
Image

Quantum Computing

AI To Control Quantum Computers

Quantum computers are able to solve complex calculations, but the quantum states are sensitive to constant interference from its environment. Researchers at Max Planck Institute for the Science of Light are using neural networks to develop error-correction learning in order to provide “active protection” for a quantum computer. The team has over 2000 artificial neurons which uses AlphaGo-inspired architecture to learn how to perform a task which will be essential for operation of future quantum computers. This process is known as quantum error correction. With continuous training of the algorithm, it will get better over other error-correction strategies. Read More

Image
Image

Quantum Computing

Quantum Computers Proved to Be More Powerful than Classical Computers

International team of researchers have collaborated at Technical University of Munich and provided a conclusive proof that quantum computers have computational advantage over classical computers. They designed a quantum circuit which solved a particular math problem that would had been impossible for a classical computer to solve, subject to same constraints. The advantage was due to “non-locality”; a feature of spatially isolated quantum system that allows them to be considered as single system. Design of quantum circuits, has a tradeoff between number of qubits interacting in the circuit and the number of operations that can be performed. This was overcome by designing quantum circuit in which several shallow circuits operate in parallel, acting as single system due to nonlocality. This allows quantum computers to perform calculations faster than a classical computer. Read More

Image
Image

Decimal Point Analytics Pvt Ltd does not make any recommendation, solicitation, or offer for any securities and is not responsible for suitability of any securities for any purpose, investment or otherwise. It is the sole responsibility of the client, as a professional organization, to exercise professional due diligence in ensuring suitability of investment and ensuring that when the client publishes a part or full report under its own brand, the legal requirements for distribution of such material are complied with in all the jurisdiction in which it is published. Decimal Point Analytics Pvt. Ltd. shall not be responsible for any loss suffered by the user. The returns indicated, including future projections, in any investment report prepared by Decimal Point Analytics Pvt Ltd are not guaranteed in any manner and may not be achieved.

India :

5A, B-Wing, Trade Star Building, J. B. Nagar, Andheri-Kurla Road, Andheri (East), Mumbai - 400 059, Maharashtra, India
+91 22 30015200
info@decimalpointanalytics.com

United Kingdom:

1st Floor, 99, Bishopsgate, London, EC2M 3XD, United Kingdom

+ 44 20 3286 1998
info@decimalpointanalytics.com

USA:

17 State Street, Suite 4000, New York, NY 10004 U.S.A

+1 (917) 341 3218
info@decimalpointanalytics.com

www.decimalpointanalytics.com