Tech update featuring Facebook, Google and LinkedIn

Facebook open-sources DLRM, a deep learning recommendation model

Global Social media giant, Facebook has revealed the open source release of Deep Learning Recommendation Model (DLRM), a state-of-the-art AI model for serving personalised results in production environment. DLRM can be found on GitHub and its implementations are available for Facebook’s PyTorch, Facebook’s distributed learning framework Caffe2 and Glow C++.
Facebook’s AI Research (FAIR) is creating DLRM to enable the wider AI community to address challenges presented by recommendation engines. The makers of DLRM recommend that the model be used for benchmarking the speed and accuracy performance of recommendation engines. The benchmark for experimentation and performance evaluation is written in Python and supports random synthetic inputs.

Google AI Team uses Mannequin Challenge Videos to Train AI

Search engine giant, Google’s AI team is using 2,000 Mannequin Challenge videos to train a neural network to understand 3D scenes, it is basically training AI to predict depth in the videos. While humans are naturally good at interpreting 2D videos as 3D scenes, machines need to be taught and the team is honing the neural network’s ability to reconstruct depth of freely moving objects, which will help the robots manoeuvre in unfamiliar surroundings. For the training programme, 2,000 videos were converted into 2D images with high-resolution depth data which made it easier for the network to predict the depth of the moving object at a much higher accuracy than it used to earlier.

Google debuts Deep Learning Containers in beta

Google has rolled out a beta availability of a new cloud service that offers environments optimized for deploying and testing applications powered by deep learning (AI that mimics the human brain’s ability to tackle problems). This service, Deep Learning Containers can be run in the cloud or on-premises. It also consists of performance-optimized Docker containers with various tools to run deep learning algorithms.
The service also provides machine learning acceleration capabilities with Nvidia Corp.’s graphics processing units and Intel Corp.’s central processing units. It also supports PyTorch and TensorFlow. It can also be hosted on Google’s Compute Engine and Kubernetes Engine services or on the Google AI platform. With Deep Learning Containers, Google is making machine learning environments easier for developers to set up and receive access.

LinkedIn lifts the hood on its news feed algorithm to show how it ranks posts

Business and employment-oriented platform, LinkedIn has announced several changes to its ranking algorithm. The company is moving on from ranking trending content towards prioritising niche-specific professional conversations. To determine posts that receive a higher ranking in someone’s feed, LinkedIn’s algorithm uses AI to identify niche, occupationspecific conversation, for instance there is more conversation around #performancemanagement (niche) as compared to #management (broad). If a connection uses a hashtag and you follow it, the post gets an extra boost. By revealing this algorithm, LinkedIn is giving marketers insights on how to utilize the platform optimally.

Salesforce’s AI grasps commonsense reasoning

CRM and cloud computing services provider, Salesforce has proposed a new open source corpus – Common Sense Explanations (Cos-E) for training and interference with a novel machine learning framework (Commonsense Auto-Generated Explanation, or CAGE) which improves performance on question and answer benchmarks by 10% over baselines. The commonsense explanations for CoS-E is divided in two parts- a question split and a random split. For the commonsense reasoning model — a classification module that learned to perform predictions on the (Common sense Question Answering) CQA task — the team chose Google’s BERT, which is unique in that it’s both bidirectional (allowing it to access context from past and future directions) and unsupervised (meaning it can ingest data that’s neither classified nor labeled).

ADDA launches new AI initiative in partnership with IBM

US-based tech giant IBM in collaboration with the Abu Dhabi Digital Authority (ADDA) has launched a set of AI training workshop for the Abu Dhabi government employees. These workshops will help to increase awareness on benefits offered by AI, while also improving the decision making skill of the government employees and to develop all sectors of the emirate. The workshops will cover topics including business management and technical aspects of modern technologies and will be held in batches; the first batch will start will 13 government entities.

IBM unveils new Data Prep Tool Designed to help speed DataOps

IBM has rolled out InfoSphere Advanced Data Preparation, a solution designed to assist clients in transforming raw data sets by formatting, structuring and enriching them for analytic processing and reporting. Jointly developed with Trifacta, the new InfoSphere is designed to work with clients’ existing data environments, data lakes included.
Among many features, InfoSphere has an intuitive dashboard for visualising the data prep process, including the progress of tracking data quality and lineage. Residing on top of a client’s data lake or warehouse, the solution provides automated transformation capabilities. The tool is also designed to empower users of all levels of technical expertise to generate data insights.

Oracle expands analytics offerings to aid customer experience

Cloud Oracle’s newly launched Oracle Analytics Cloud Platform handles business processes and offers augmented analytics, natural language processing, cloud computing, machine learning, enterprise planning workflows and data-analysis services. Oracle’s Analytics Cloud Platform, which includes Oracle Mobile Cloud Enterprise, Oracle Data Integration Platform Cloud, and Oracle API Platform Cloud will help business improve as well as reduce risk.
Users will profit from the enhanced performance, favourable cost, reliable security and innovation offered by these solutions. Oracle’s analytic abilities are available in the cloud via Oracle Analytics Cloud while on-premises is accessible via Oracle Analytics Server. This addition to its cloud platform reflects Oracle’s expansion of cloud computing product portfolio inorganically.

Oracle launches AML tool for small banks

Oracle Financial Services has announced the launch of its Anti-Money Laundering (AML) express edition for small and medium sized financial institutions, connecting these banks with a streamlined platform for money laundering and terrorist funding detection, investigation and reporting solutions. This solution supports affordable and straightforward compliance solutions to these smaller financial institutions. It supports faster deployment for on-premise or cloud infrastructures and has a built-in library of potential scenarios that address the common laundering behaviours for faster detection while including case management abilities.

Analytics firm SAS plans to hitch AI to real economy

SAS has plans to invest $146 million in AI with China as its major target market. Part of the investment will be used into SAS’ research and development center in Beijing whereas the rest will be used to bridge the talent gap. SAS plans to invest in local universities to help teach analytics to the next generation. SAS established a financial incubation center in Nanjing, Jiangsu province, in East China, integrating its financial system into the local ecosystem. It also set up an internet of things partnership with the local government of Wuxi, Jiangsu province, using its technology to energize the city’s outdated industries.

Global payments provider Visa is set to acquire Verifi, a payment dispute resolution technology

that helps firms reduce chargebacks. The financial terms are not yet disclosed. Verifi’s technology connects all parties in the dispute management process in real-time and aims to resolve them before they become a chargeback. By integrating Verifi’s chargeback tools into Visa, it can provide buyers and sellers more automation, real-time communication and data-driven insights.

AI-backed fintech startup Recko raises $1m in seed funding from Prime Ventures

Artificial intelligence-driven fintech startup Recko bagged $1 million in a seed funding round from Prime Venture Partners. The India-based company will use the investment to expand its team, scale across verticals and build out more capabilities of its AI technology. The company builds machine learning models to identify anomalies, risk and intelligence around money flow.

[“source=financialdirector”]