Analytics@

Introduction Core Idea Key points in Alpha-beta Pruning Working of Alpha-beta Pruning Codes in Python Move Ordering in Pruning Rules to find Good ordering Reference Introduction The word ‘pruning’ means cutting down branches and leaves. In data science pruning is a much used term which refers to post and pre-pruning in decision trees and random […]

The post Alpha Beta Pruning in AI appeared first on GreatLearning.

Source: Alpha Beta Pruning in AI by administrator

2017 holds a considerable amount of promise for big data technologies, which are branching out and merging with other cutting edge applications for artificial intelligence, the Internet of Things, and blockchain. Jelani Harper details these trends and more for big data in The New Face of Big Data: AI, Iot, and Blockchain.

Source by jelaniharper

The various elements of Artificial Intelligence have become all but commonplace to the contemporary enterprise. Facets of machine learning, Natural Language Processing, chatbots, and virtual assistants dominate numerous backend processes, providing highly desired automation.

What’s far less pervasive today is embedding these same capabilities in core business processes, where the acceleration and automation of AI directly results in greater revenues. Transforming the way individual business units function with AI usually involves a paradigm in which organizations deliver several years of current and historical data to vendors, who then create individual models for specific tasks like reducing churn.

According to Vizru Chief Executive Officer Ramesh Mahalingam, there’s a fundamental problem when organizations: “send data out, and you basically start leaking so much information that nobody is in control of the system.”

The two reoccurring dangers of this means of implementing AI include harvesting data—when vendors replicate, segment, and sell organizations’ proprietary data for the former’s benefit—and distributing malware. Both squander valuable enterprise resources, take place with alarming frequency, and are checked by user-friendly platforms giving organizations the tools to develop AI on their own.

Data Harvesting

There are manifold dimensions to the harm stemming from data harvesting with this typical method for fostering enterprise-scale AI. This practice not only exploits organizations’ proprietary data in a means similar to that of conventional data breaches, but also gives the competitive advantage such data affords to their competitors. Moreover, data harvesting produces this effect by profiting the AI vendor selling this data, not the organization whose data they rightfully are. “Once 15 years of data is handed down, what does the vendor do with that data?” commented Mahalingam. “That’s what we mean by harvesting. There is so much information that you can slice and dice, you can send it to different models for yourself, anonymize it, or otherwise. You can actually sell that data to competitors in so many different ways.”

In most instances, it’s almost impossible for organizations to realize if their data has actually been harvested and leveraged by vendors. For example, data in the financial services industry can be sold to an organization’s competitors, to analysts following certain developments, or manufacturers gaining unparalleled insight into market trends based on this information. All of those profits should go to the organization that initially collected that data; with many contemporary AI vendor practices, it simply goes to the latter.

Distributing Malware

Data harvesting implies organizations don’t know who else is capitalizing on their data. Distributing malware implies organizations don’t know exactly what they’re getting when their data are returned—or when they implement solutions devised by vendors based on that data. This concern is one of the fundamental reasons organizations remain skeptical of just giving their data to third-party AI vendors. Once an organization’s data are outside the watchful gaze of IT teams and cyber security personnel, there are no guarantees those datasets will remain protected or follow data governance protocols.

“Some of the largest banks, some of the largest insurance companies, they all worry about companies harvesting data and becoming a malware [distributor] and they not knowing about it,” Mahalingam mentioned. “Because, IT’s not making decisions on its own anymore, line of business runs it, and line of business just thinks it’s just some fast point solution to do something small.”

It’s a distinct possibility organizations can have their data returned with malware. In this instance, the AI vendor is the initial malware distributor, but whoever interacts with that data going forward—partners, contractors, different business units, etc.—can be potentially exposed to malware as well. When accessing third-party AI vendors for virtually any service like text analytics, for example, organizations can receive their data and services in a way which “they send you back a file,” Mahalingam said. “That information that comes back to your system can turn into malware. It can infect the rest of your environment.”

Taking Precautions

When accessing enterprise AI solutions through third-party vendors, organizations run the risk of encountering various aspects of data harvesting and malware distributions. The former enables others to capitalize on an organization’s data; the latter can severely compromise productivity for organizations and their partners by causing security and compliance issues like ransomware. The danger about both of these situations is “when you have 170,000 companies providing services to the market, it is impossible for you to go and do due diligence on all of these companies,” Mahalingam reflected. “Rather, you need to bring control within your environment.”

Organizations can accomplish this objective by accessing AI services through platforms designed for in-house AI for non-technical, citizen data scientists. Competitive solutions in this space utilize a stateful network for processing AI that serves as a guardrail for accessing third-party services, solidifying the security and governance concerns forsaken in the conventional enterprise AI paradigm. With this approach, data harvesting and distributing malware is mitigated, giving organizations more control over their data and AI resources.

Source by jelaniharper

Introduction

The scale and rapid pace of technological change has spawned entirely new business models and ways of going to market. And as value creation strategies change, so too must the traditional finance benchmarks associated with value creation.

Nearly 90 percent of CFOs surveyed during Oracle’s September 2013 CFO Summit agreed that the use of emerging technologies is leading them to reconsider or introduce new finance best practices in key areas such as cloud and mobile adoption to deliver insights and innovative new functionality to employees, predictive analytics and big data to improve planning and forecasting, and business support metrics to identify how finance can better support strategic lines of business.

While traditional finance benchmarks such as how fast a company can close its books or reduce its days sales outstanding (DSO) rate still matter, new finance benchmarks increasingly are top of mind for CFOs. For example, Deloitte reported in its Q4 2013 CFO Signals survey of North American CFOs that business support metrics are the most important finance measure today¹. Traditional measures including head count, finance costs as a percentage of revenue, and velocity of reporting still rank highly but there has been a notable shift toward business support metrics. These can include internal client satisfaction scores that can shine a light on the success of new finance best practices such as business partnering.

More broadly, the potential impacts on finance best practice are as varied as the technologies themselves, but the ability to drive more economic value from data-driven insights coupled with the agility conferred by cloud and mobile computing are standout areas.

The embedding of business intelligence and predictive analytics in transaction systems, combined with hardware that is 20 times more powerful than even a year ago, allows the finance function to churn large volumes of data at levels of granularity that were previously unattainable. This rich capability has been game-changing, allowing the finance function to use real-time data to quickly respond to market volatility, spot trends, more accurately predict outcomes, and proactively drive top-line growth as well as contain or reduce costs. Best practice reflects a shift in emphasis from traditional reporting on historic data to partnering with other line-of-business functions to proactively influence the drivers of growth and cost reduction.

New technology is also causing CFOs to re-examine the traditional yardsticks of investment appraisal. Payback, discounted cash flow, and return on capital employed become blunt instruments in the new era of big data and social business where the link between revenues generated from structured (big data) and unstructured data (social media) becomes more tenuous. For example, a McKinsey social media survey² revealed that while more than 70 percent of companies believe that digital marketing holds significant potential, more than half struggle to measure its exact impact on sales and profits. CFOs certainly have their work cut out to establish best-practice measures in this rapidly evolving environment.

The cloud is delivering greater operational efficiency through automation of finance processes where traditional systems simply cannot cope. But the cloud is also exposing the inadequacies of traditional ways of measuring the total cost of ownership (TCO) of systems as CFOs grapple with the implications of monthly subscription fees in place of perpetual software licenses and annual maintenance fees.

The cloud is even reshaping finance best practices around the buying decision as organizations struggle to compare the ROI of cloud versus on-premises propositions and the profoundly different operating model (for example, infrastructure, upgrades, head count) that can result from cloud deployment. Substituting capital expenditure with operating expenditure can have a material effect on cash flow and taxation and possibly budget allocations in some public sector bodies. Even where the transactions are notionally located in the cloud can give rise to unexpected tax challenges.

According to an Oracle CFO webcast with the FEI in 2013, organizations seeking to capitalize on the transformative qualities of mobile computing are actively encouraging a “mobile first” policy for applications such as mobile business intelligence and dashboards. But the rapidly rising popularity of mobile technology is also causing CFOs to quickly reevaluate finance best practice around where mobile computing can add value, as well as which employees and applications should be supported and provided access. Furthermore, as mobile apps proliferate, CFOs are scrambling to codify policy and procedure around security and confidentiality. The “bring your own device” (BYOD) phenomenon is adding to the strain as CFOs seek to guard against the consequences of stolen devices with corporate data on them while at the same time developing fair compensation packages for employees using their own mobile phones and tablet devices in the workplace.

Of course, the rapid evolution of best practice also places heavy demands on finance talent. Accenture strategy consultant David Axson recently stated, “The success of the CFO will be measured by the success of the organization as a whole.” CFOs at the top of their game are uniquely placed not only to influence the success of their own domain but also to secure the future success of their organization, and the most competitive CFOs are using emerging technologies and new finance best practices to benchmark success and deliver on their goals. To help CFOs adopt new finance best practices based on today’s emerging technology imperatives, Oracle has partnered with Accenture to produce a new series entitled Five Minutes on Modern Finance. You can find the first issue here.

Note: This article originally appeared in Oracle. Click for link here.

Originally Posted at: Finance Best Practices Are Changing—Is Your Organization Keeping Pace? by analyticsweekpick

Artificial Intelligence (AI) systems that companies claim can “read” facial expressions is based on outdated science and risks being unreliable and discriminatory, one of the world’s leading experts on the psychology of emotion has warned.

Lisa Feldman Barrett, professor of psychology at Northeastern University, said that such technologies appear to disregard a growing body of evidence undermining the notion that the basic facial expressions are universal across cultures. As a result, such technologies – some of which are already being deployed in real-world settings – run the risk of being unreliable or discriminatory, she said.

“I don’t know how companies can continue to justify what they’re doing when it’s really clear what the evidence is,” she said. “There are some companies that just continue to claim things that can’t possibly be true.”

The AI system, developed by the company HireVue, scans candidates’ facial expressions, body language and word choice and cross-references them with traits that considered to be correlated with job success.

Amazon claims its own facial recognition system, Rekognition, can detect seven basic emotions – happiness, sadness, anger, surprise, disgust, calmness and confusion. The EU is reported to be trialling software which purportedly can detect deception through an analysis of micro-expressions in an attempt to bolster border security.

“Based on the published scientific evidence, our judgment is that [these technologies] shouldn’t be rolled out and used to make consequential decisions about people’s lives,” said Feldman Barrett.

However, a growing body of evidence has shown that beyond these basic stereotypes there is a huge range in how people express emotion, both across and within cultures.

In western cultures, for instance, people have been found to scowl only about 30% of the time when they’re angry, she said, meaning they move their faces in other ways about 70% of the time.

“There is low reliability,” Feldman Barrett said. “And people often scowl when they’re not angry. That’s what we’d call low specificity. People scowl when they’re concentrating really hard, when you tell a bad joke, when they have gas.”

The expression that is supposed to be universal for fear is the supposed stereotype for a threat or anger face in Malaysia, she said. There are also wide variations within cultures in terms of how people express emotions, while context such as body language and who a person is talking to is critical.

“AI is largely being trained on the assumption that everyone expresses emotion in the same way,” she said. “There’s very powerful technology being used to answer very simplistic questions.”

Source: The Guardian

Originally Posted at: AI systems claiming to ‘read’ emotions pose discrimination risks by administrator

Metadata has long been considered a critical component of both data management and data governance for its ability to provide apt descriptions of data. According to Cambridge Semantics Chief Technology Officer Sean Martin, however, metadata management is also essential because “the processing to move [data] around and make use of it also needs to be described as metadata.”

By centralizing metadata management with cataloguing techniques across the enterprise, organizations attain a degree of flexibility in their movement of data that’s difficult to duplicate. Oftentimes, metadata descriptions are the first step in automating action for business objectives as varied as arranging diverse data into a single regulatory accord or dynamically switching cloud providers to minimize costs.

The result is an improved capacity for data governance and business processing that maximizes data’s automation capabilities so that they “become less brittle and less manually set up, configured and run,” Martin said.

Metadata Cataloguing
Because metadata management describes both data elements and data’s movement, metadata catalogues provide comprehensive information about most aspects of data. That information pertains to everything from original source data location and schema to relevant mapping for increased business understanding of data’s meaning. By harmonizing data with standardized models, organizations can also readily trace data’s provenance as needed. After initially ingesting data, “in the catalogue you’ve got an item that says here’s a dataset that contains data with these concepts, this is the ontology that describes it, here’s some metadata about it,” Martin commented. “It includes all the details that you need to locate that data and the user of that data.” Those details are also relevant to the processing required of data-driven action. The more granular those metadata descriptions are about that action, the greater the propensity for automation. Common processes such as issuing analytic queries or running jobs for applications are denoted with metadata. “You’re describing the processing you might want,” Martin said. “Configurations of cloud computing on-the-fly using cloud APIs and automation. But you need abstract metadata descriptions to describe that automation for doing things like actually, literally, moving the data.”

Describing Automation
The meticulous maintenance of data catalogues via metadata descriptions of the data and their processing expedites action. Descriptions can include code for transformation or analytics, so that the resulting action is largely automated. The specificity of the descriptions is as precise as the action desired, and might encompass “which version of the software to use, how many nodes, how much data there is so we know how big the compute nodes need to be, and how much RAM they need to have,” Martin revealed. “There’s that kind of information, as well as what data needs to be loaded and what ELT rules need to be run.” By compiling all of this information into a data catalogue, users have substantially less to do when creating data-generated action. “All of that is a description so when some user says I want that, it passes into the system and the workflow manager picks it up and actually runs the description to do all that work using the configuration,” Martin explained.

Impacting Data Governance
The automation benefits of the more responsive data-driven action of effective metadata management naturally lend themselves to multiple facets of data governance. In particular, the approach described by Martin is useful for demonstrating regulatory compliance—which is one of the fundamental drivers for data governance programs across industries. By quickly moving data around, companies can align data for requirements as they arise. “The metadata can be used for all sorts of things including special processing,” Martin said. “You know, I need this processing to happen over here because it’s cheaper or because this data must never leave the country. Privacy and the European privacy laws that don’t allow data to escape from Europe might be another issue.” The same concept applies to aspects of security in instances in which data are not allowed to leave company firewalls. Additional issues of regulatory compliance could require shuffling data around in certain formats to demonstrate adherence. “Regulations are some of the drivers for the banks and pharmaceutical companies to put more and more of their data in an organized form simply to survive audits by regulators,” Martin observed.

Shifting Landscape
As the impact of big data continues to reverberate throughout the industry, the data landscape itself continues to shift while organizations struggle to keep pace. Technologies have changed, approaches have come and gone, and applications have become more sophisticated. What has yet to change—and quite likely will not—is the value of metadata in providing timely management of data assets. Effectively managing that metadata with a centralized catalogue describing both the data and its processing is the key to enabling automation and other impending necessities of tomorrow’s data landscape. Metadata management is also vital to implementing sustainable data governance and increasing agility at a time when it’s needed most.

Source by jelaniharper