<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Sasha Lazarevic on Medium]]></title>
        <description><![CDATA[Stories by Sasha Lazarevic on Medium]]></description>
        <link>https://medium.com/@sasha.lazarevic?source=rss-ad51a0e4e895------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*6cwi3TuC_dcnY14P.</url>
            <title>Stories by Sasha Lazarevic on Medium</title>
            <link>https://medium.com/@sasha.lazarevic?source=rss-ad51a0e4e895------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 21:31:54 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@sasha.lazarevic/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Увод у квантно рачунарство]]></title>
            <link>https://medium.com/quantum-serbia/introduction-to-quantum-computing-serbian-3185a23c0cd4?source=rss-ad51a0e4e895------2</link>
            <guid isPermaLink="false">https://medium.com/p/3185a23c0cd4</guid>
            <category><![CDATA[quantum-computing]]></category>
            <category><![CDATA[kvantno-racunarstvo]]></category>
            <category><![CDATA[квантно-рачунарство]]></category>
            <dc:creator><![CDATA[Sasha Lazarevic]]></dc:creator>
            <pubDate>Mon, 06 Dec 2021 11:24:18 GMT</pubDate>
            <atom:updated>2021-12-06T11:24:18.844Z</atom:updated>
            <content:encoded><![CDATA[<p>Шта су квантни рачунари, како функционишу и која је њихова предност у односу на класичне (или ”не-квантне” рачунаре) ?</p><p>У овој серији чланака желим да објасним неке од најважнијих постулата и чињеница везаних за квантно рачунарство и квантне технологије, па ако вас занима ова горућа тема, наставите са читањем.</p><h3>”Квантна” историја</h3><p>Пре више од сто година група еминентних светских научника открила је и описала квантне феномене, па мислим да је на почетку ове серије вредно поменути њихова имена и доприносе: Прво, Макс Планк (Max Planck) је 1900-те говорио о фотонским „квантима“ у смислу да светлосна енергија може постојати само у дискретним пакетима, а ово је навело Нилса Бора (Niels Bohr) 1913-те да примени исти принцип и на енергију електрона. Даље, Ервин Шредингер (Erwin Schrödinger) и Вернер Хајзенберг (Werner Heisenberg) су 1926-те независно развили два обједињујућа математичка модела атомских квантних феномена као што су суперпозиција, неодређеност, сплетеност и проблем квантних мерења. Исте године Енрико Ферми (Enrico Fermi) је описао дистрибуцију честица по енергетским стањима, а Волфганг Паули (Wolfgang Pauli) је 1927. идентификовао како честице интерагују са спољашњим електромагнетним пољем. Такође 1927. године, Пол Дирак (Paul Dirac) је створио квантну теорију поља. Занимљиво је да Алберт Ајнштајн уопште није веровао у недетерминистичку природу квантних појава и целу теорију је комплетно одбацио. Што се тиче Николе Тесле, он се бавио пољима енергије, али није улазио у њихову квантну природу. Ето, то су у основи пра-очеви квантне теорије који су оставили веома богато наслеђе, данас препознато кроз концепте и термине као што су Планкова константа, Хајзенбергов принцип неизвесности, Шредингерова једначина, Диракова нотација, Паулијеве матрице и друге. Они који желе да проучавају ову тему дубље, дефинитивно ће срести ова имена и по њима названу терминологију, а такође ће научити и о Хилбертовом простору, названом по математичару Давиду Хилберту (David Hilbert), и Блох сфери названој по Феликсу Блоху (Felix Bloch), два научника који су такође значајно допринели овом раном развоју квантне теорије и квантне механике.</p><p>Након овог почетног периода, физичари су углавном радили на експериментима како би доказали валидност постулата квантне механике и коначно су 2015. разрешили све сумње када су три независна експеримента показала да не постоји друго објашњење за квантно понашање честица попут фотона и електрона, осим од оних које пружа квантна теорија.</p><p>Квантно рачунарство у облику квантне информационе науке (quantum information science) је рођено крајем 1970-их. Најутицајнији истраживачи који су помогли у развоју овог теоријског концепта били су Пол Бениоф (Paul Benioff), Јуриј Мањин (Юрий Иванович Манин), Ричард Фeјнман (Richard Feynmann) и Дејвид Дојч (David Deutsch). Њихова главна идеја је била да се квантни феномени могу и требају искористити за обраду информација. Фејнман је постао познат 1982-ге године по теорији и предлогу да се може направити релативно једноставни, вештачки квантни систем који би се користио за симулацију других, природних и много сложенијих, квантних система.</p><p>Али можда је најзанимљивији период почео 1990-их са развојем алгоритама, који су подстакли велико интересовање и улагања у квантно рачунарство:</p><ul><li>Дојч-Џоза (Deutsch-Jozsa) алгоритам, 1992-ге</li><li>Шоров (Schor’s) алгоритам 1994-те (познат по томе што може да дешифрује криптоване податке)</li><li>Гроверов (Grover’s) алгоритам из 1996-те за претраживање база података</li></ul><p>Што се тиче хардвера, први квантни рачунари створени су 1998-ме, истовремено у три лабораторије: на Оксфорду, IBM-овој лабораторији у Алмадену (Калифорнија) и Берклију. Ове машине су биле засноване на нуклеарној магнетној резонанци, и били су коришћени за екпериментисање са једно-кубитним системима, али је ова технологија касније напуштена због недостатка могућности да се више квантних система међусобно сплету (феномен познат као сплетеност / entanglement) и прилично значајног квантног шума. Први пут се суперпроводнички квантни рачунар наводи од стране јапанских истраживача, а онда је направљен и експериментално 1999. године од стране IBM-a. IBM има дугу историју истраживања у области суперпроводничких кола која датира из 1960-их, тако да им је било прилично лако искористити то искуство да би развили специфичне суперпроводнике баш за квантне рачунаре. Током времена су ова суперпроводничка кола значајно еволуирала и тренутно су веома обећавајућа технологија и може се рећи state-of-the-art за извршавање квантних алгоритама.</p><p>Од 2016-те је могуће извршавати инструкције на IBM-овим квантним рачунарима који су <a href="https://quantum-computing.ibm.com/">јавно расположиви</a> , а данас је та понуда још богатија са компјутерима фирми као што су Google, Rigetti, IONQ, Honeywell, Alibaba, Xanadu, као и бројним другим који се развијају у истраживачким лабораторијама широм света, и који поред суперпроводничких кола користе и јонске замке, фотонске процесоре, квантне тачке или хладне атоме.</p><p>Ово је била кратка ”квантна” историја, а у следећем чланку ћу писати о основи квантног рачунарства : квантним стањима и кјубитима. У међувремену, препоручујем и чланак са <a href="https://sr.wikipedia.org/wiki/%D0%9A%D0%B2%D0%B0%D0%BD%D1%82%D0%BD%D0%B0_%D0%BC%D0%B5%D1%85%D0%B0%D0%BD%D0%B8%D0%BA%D0%B0">Википедије о квантној механици</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3185a23c0cd4" width="1" height="1" alt=""><hr><p><a href="https://medium.com/quantum-serbia/introduction-to-quantum-computing-serbian-3185a23c0cd4">Увод у квантно рачунарство</a> was originally published in <a href="https://medium.com/quantum-serbia">Quantum Serbia</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Think 2019 — Part 5]]></title>
            <link>https://medium.com/@sasha.lazarevic/think-2019-part-5-2de15b83adfe?source=rss-ad51a0e4e895------2</link>
            <guid isPermaLink="false">https://medium.com/p/2de15b83adfe</guid>
            <category><![CDATA[ibm]]></category>
            <category><![CDATA[ibm-cloud-garage]]></category>
            <category><![CDATA[think2019]]></category>
            <dc:creator><![CDATA[Sasha Lazarevic]]></dc:creator>
            <pubDate>Tue, 26 Feb 2019 13:35:34 GMT</pubDate>
            <atom:updated>2019-02-26T17:23:58.673Z</atom:updated>
            <content:encoded><![CDATA[<h3>Think 2019 — Part 5</h3><p>In previous parts of my Think 2019 report, I wrote about AI, Cloud, Infrastructure, Blockchain and Applications. Here is the conclusion of the event and some other, not less important notes, like IBM Garage program external partnerships, education programs, signed deals and IBM investments.</p><h3>Other Domains</h3><p>Not less important, but simply I was not able to follow everything. I’ve hear about interesting <strong>IOT</strong> use cases at KONE elevator company and L’Oreal, and synergies with machine learning algorithms to streamline the manufacturing and operations, <strong>Security</strong> with QRadar products, <strong>Networking</strong> around Cisco and VMWare NSX-T</p><p>I have listed here below what words were the most used at Think 2019, This is my personal perception, maybe some other participants will disagree with it.</p><ol><li>AI (including Watson and Data)</li><li>Kubernetes</li><li>Hybrid</li><li>Quantum</li><li>Microservices (including DevOps)</li><li>Multicloud</li><li>Blockchain</li><li>Security</li><li>Power</li><li>Apps</li></ol><h3>Partners</h3><p>A lot of exhibitors, and companies that work together with IBM.</p><p>The most prominent was of course <strong>Red Hat</strong>. No new information about pending merger, Red Hat executives were talking about hybrid cloud, Linux, OpenShift, Kubernetes, and other technical topics.</p><p><strong>VMWare</strong> was talking about hybrid cloud as definitely the way how the compnting will work in the future, due to laws of physics (latency), laws of economics (where it is cheaper) and laws of countries (where you need to keep your data). I mentioned about IC4V and other interesting technologies for the hybrid cloud.</p><p><strong>ServiceNow</strong> had an interesting presentation on how to deal with complexity of IT Management tasks. Most of organizations have multiple tools for infra and app management. Why not have something like Tools-as-a-Service , a cloud with these tools where the onboarding of new systems will be simplified. Based of course on ServiceNow. Interesting idea, hopefully this will translate into more concrete solutions.</p><p><strong>ScienceLogic</strong> has a very important role in the world of multicloud, and partner with IBM for the monitoring and event management solutions. They have created toolset for AIOps based on data lake, and they support containers.</p><p><strong>Cisco</strong> is specialist for multicloud. There was a presentation on security where IBM QRadar imports threat related data from Cisco devices</p><p>Chinese company <strong>Inspur</strong> (large technology provider, I wrote about them back in 2014) presented servers based on OpenPower architecture, jointly developed with IBM.</p><p>I didn’t see <strong>Lenovo</strong>, which is strange for me. I don’t know why they were not more visible.</p><h3>Business Partner Program</h3><p>This was an important topic. IBM has deployed a new platform for the business partners based on Watson to match the partners with clients, and other partners, and with IBM offerings. This should streamline go-to-market approach.</p><p>We’ve heard that IBM needs industry-specific, and domain-specific software providers to run their applications on ICP4Data and collaborate with IBM.</p><h3>Support Organization and Education</h3><p>It was interesting to speak with people involved in these teams and learn about how IBM wants to help clients modernize and adapt to this new, cloud and AI-driven world</p><h3>Garages</h3><p><strong>IBM Cloud Garage</strong> : this geographically distributed team organizes design thinking workshops, architecture workshops, solution build-ups and helps the clients from discovery, design, to development and deployment. These guys and girls accompany the businesses throughout the whole cloud journey. They use IBM Cloud Garage Method — agile approach to build applications and solutions with cloud in mind.</p><p>There are IBM Cloud Garages in multiple locations: San Francisco, New York, Toronto, Melbourne, London, Madrid, Nice, Copenhagen, Münich, Sao Paolo, Singapore and Tokyo. Interesting concept also of pop-up garage where the team can come to physically closer or customer location for some limited time.</p><p><strong>IBM Garage</strong> : it has larger scope. It is based on collaboration between different teams within IBM and can include AI, Quantum, Blockchain and other solution lines. But the underlying method is the same.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/889/0*SRuU3bOyOZ8RP_9N.png" /></figure><p>IBM Studio exists also in many locations (one of them is Zürich in Switzerland). The only difference is that it normally doesn’t cover the development phase of the digital transformation.</p><h3>Education</h3><p>IBM announced <strong>IBM AI Certification and Learning program</strong>. This education program focuses on training and certifying professionals of our customers, not only in technical expertise, but how to evaluate business potential of an idea, design a solution based on microservices and AI, and implemented it using open-source and other IBM frameworks.</p><p>I spoke also with people from <strong>IBM Global University Program</strong>, partnership between IBM and some well-known universities. The idea is to alleviate the problem of shortage of skilled professionals through blended on-line and classroom programs in different domains like data and AI, Blockchain and others. Check <a href="http://research.ibm.com/university/">here</a> for more information.</p><p>Certification was also organized during the conference. Pearson Vue gave a discount to those who wanted to test their knowledge in one of IBM technologies. I prepared and passed the exam <strong>Watson Certified Application Developer</strong>. I’ve seen quite a lot of people in the room concentrated in front of computer screens. Please check this page to prepare for your own exam: <a href="https://www.ibm.com/certify/">https://www.ibm.com/certify/</a></p><h3>Deals</h3><p>Few deals and investments worth mentioning:</p><ul><li>5-year contract with Santander worth $700M for the busienss and digital transformation of this Spanish bank towards hybrid cloud, AI for customer experience, big data for productivity and security,</li><li>Deal with ICBC Argentina worth $63M to transform customer services using AI and Cloud. This is Argentinian subsidiary of the biggest bank in the world, Chinese ICBC</li><li>I would highlight here also an announcement from the week before Think, the investment of $2B in a new AI research lab in New York for the computer-chip R&amp;D</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CuuUSfbyj9ONS0PrRI5qmg.jpeg" /></figure><h3>Conclusion</h3><p>Tech event of the year, Think 2019. I was able to meet very interesting people, learn a lot , share my knowledge, get certified, and above all, get inspired. I saw good organization and good atmosphere (despite the rain). Now I understand better where this world of technology is heading, and where IBM and our customers are going. There are quite a lot of challenges on that road, and people are those who will resolve them. And people who are inspired and who Think are a very powerful force.</p><p>Think (about) 2019 !</p><p>Sasha Lazarevic, Switzerland</p><p>Also published in my personal blog: <a href="http://lzrvc.com/"><strong>https://LZRVC.com</strong></a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aQ453TOp98h9N7l2qcgtpg.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2de15b83adfe" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Think 2019 — Part 4]]></title>
            <link>https://medium.com/@sasha.lazarevic/think-2019-part-4-a2185c949fad?source=rss-ad51a0e4e895------2</link>
            <guid isPermaLink="false">https://medium.com/p/a2185c949fad</guid>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[application-modernization]]></category>
            <dc:creator><![CDATA[Sasha Lazarevic]]></dc:creator>
            <pubDate>Tue, 26 Feb 2019 13:28:38 GMT</pubDate>
            <atom:updated>2019-02-27T17:24:16.749Z</atom:updated>
            <content:encoded><![CDATA[<h3>Think 2019 — Part 4</h3><p>In <a href="https://medium.com/@sasha.lazarevic/think-2019-part-3-4c8e8bba8008"><strong>Part Three</strong></a> I wrote about the infrastructure news and announcements from Think 2019. In this Part 4, I’m sharing what I learned in the domains of Blockchain and Applications. Continue reading !</p><h3>Blockchain</h3><p>Here is another transformative and disruptive technology where IBM is very much invested with a range of projects. Despite the significant decline of cryptocurrency market, Hyperledger usage is growing and it is expected that by 2030 all combined blockchain technologies will create $3.1T of business value.</p><p>IBM announced availability of <strong>IBM Blockchain Platform 2.0 Beta</strong>. It is based on Kubernetes architecture, available also on-prem installation on top of ICP. It is also available on AWS platform. Composer is no longer there and smart contracts are developed with Visual Studio Code with SDK for Node.js, Java and Go. For more information, you can check <a href="https://console.bluemix.net/docs/services/blockchain/index.html#get-started-ibp">here</a>.</p><p>IBM has also announced <strong>Blockchain Health Utility Network</strong>. Several companies in the health industry have already joined this network, and the idea is to enable organizations to build, share and deploy blockchain-based solutions in healthcare. Two groups of use cases are planned : sharing health-related data and patients records, where the patients will be able to control with whom and how they will share their personal data. Another is related to payments and reimbursements based on smart contracts and real-time execution.</p><p><strong>IBM Food Trust</strong> <strong>Network</strong> was highlighted in one of the presentations. It is a network of companies like Walmart, Carrefour and 50+ their suppliers, based on permissioned ledger. The idea is to improve trust and transparency in the food supply chain, and reduce inefficiencies created from them.</p><p>There was an interesting application that combines AI and Blockchain: <strong>Crypto Anchor Verifier</strong> that uses visual recognition to identify authenticity of various products (already used for diamonds for example) and can write the authenticated records in blockchain. <a href="https://www.ibm.com/blogs/research/2018/05/ai-authentication-verifier/">https://www.ibm.com/blogs/research/2018/05/ai-authentication-verifier/</a></p><p>First cryptocurrency has been created on Hyperledger mainnet. It is <strong>Metacoin</strong>, actually a utility coin. No mining, no transaction fees, and hyperledger nodes are installed on LinuxONE to ensure security and reliability. A nice combination of Blockchain and Mainframe technologies.</p><p><strong>IBM Blockchain World Wire</strong> attracted a lot of attention. Jesse Lund answered to a lot of questions, but no new announcements he said, watch out on March 19 at Money 20/20 Asia in Singapore where there will be a special product announcement. IBM Blockchain World Wire is a new system for cross-border payments, where payment messaging and instructions for settlements can happen in real time. It is not Hyperledger-based, but uses Stellar platform (<a href="https://www.stellar.org/">https://www.stellar.org/)</a>. Lumens (XLM) tokens are used to transfer the value from one currency to another during the money transfer. Stellar is public permissioned network (Hyperledger is private), which was forked from Ripple and is very fast , supports thousands TPS. There are no plans to convert Lumens into security tokens.</p><p>It is interesting to see that IBM makes a step aside from Hyperledger and invests in a different platform, but it makes sense as we remember that IBM mainframes run 60% of the world transactional systems. These mainframes can easily be connected to this new international payment system, if it starts gaining traction. And this should happen, as SWIFT is expensive, slow and according to some specialists creates a lot of errors (up to 10% of transactions). Another point is that Hyperledger-based trust networks can also be connected to World Wire, as these are different but complementary technologies and use cases. Check here for more information: <a href="https://www.ibm.com/blockchain/solutions/world-wire">https://www.ibm.com/blockchain/solutions/world-wire</a></p><p>Blockchain is a promising technology that will transform our lives in the future. It brings transparency, trust and associated efficiencies. But to reap the full benefits of it, new kind of applications needs to be built and the existing systems have to be adapted to be able to interface with Blockchain.</p><h3>DevOps, Microservices and Applications</h3><p>There was quite a lot of sessions around these topics, which was expected, as the application development is actually the main battlefield. I mentioned millions of developers currently participating in Red Hat projects, but in addition to this, there is also quite a lot of developers working for IBM’s clients on the maintenance and other app improvement projects. Ideally, they should join forces and create new AI and data powered applications that will move complete industries and the whole society to the new level. This is the big idea and it was reflected in the lineup of the speakers and technologies presented.</p><p>First, featuring UrbanCode. <strong>UrbanCode Deploy</strong> is the tool used for the deployment of microservices to private cloud, public cloud, multicloud etc. There is now also UrbanCode Deploy for zSystems, enabling zDevOps. Read <a href="https://developer.ibm.com/urbancode/products/urbancode-deploy/features/urbancode-deploy-system-z/">here</a> for more details. It is tightly integrated with Kubernetes, has more than 200 plugins, many application templates and support for Terraform.</p><p><strong>IBM UrbanCode Velocity</strong> is a new component. It is enabler of DevOps analytics, where you can monitor the deployment frequency and compare it with the baselines. It allows pipeline aggregation and orchestration of complex application releases, automation (through tempaltes and calendars), so one can schedule daily and weekly application releases, and can also automatically create and delete the environments when necessary.</p><p><strong>Istio Service Mesh</strong> : This is joint and open-sourced project with Google. It helps manage, interconnect and secure microservices. Service Mesh is a network of your microservices. When you have a lot of microservices, it becomes difficult to manage how they talk to each other. This is actually where Istio comes into play. Istio is built on top of ICP and IBM announced that Istio is now available oas managed service on IBM Cloud Kubernetes service. Main features of Istio are load balancing for microservices, fine-grain control, visualizations of the microservices communications. It works through three components:</p><ul><li>Pilot (controller),</li><li>Citadel (for the encryption of the communications between microservices), which is very useful when you have service mesh over internet, in case of multiple Kubernetes clusters</li><li>Mixer (providing telemetry and logging)</li></ul><p>For more information, check <a href="https://www.ibm.com/blogs/bluemix/2018/08/five-exciting-things-istio-v1-0/">here</a></p><h3>Application Modernization</h3><p>This is a large subject and IBM talked a lot of it from different angles.</p><p><strong>IBM Cloud App Management (ICAM)</strong> : ICAM is solution for application management of microservice-based aplications. If we say that UrbanCode Deploy is rather for the Dev, ICAM is actually for the Ops part of the DevOps team. It provides end to end monitoring of applications, underlying middleware and infrastructure. With it, you can also monitor Kubernetes-managed resources. There were no new announcements, but one very good presentation on ICAM Adoption and Deployment Strategies.</p><p>ICAM has several levels : App Management Base manages traditional VMS, while App Management Advanced is adapted for Kubernetes environments on ICP and OpenShift. It requires ICP for the installation, and during the setup, one can also enable Istio Service Mesh, which becomes very important if we want to tightly control the communications between microservices or we have multicloud environment.</p><p>ICAM can show the topology of an application and its components, and you go back in time to identify eventual changes of the topology. This facilitates the discovery of changes that might have led to the performance or availability problems of the applacation. To correct the problem, one can define and use runbooks (manual or completely automated)</p><p>Next, announcement of a new product: <strong>IBM Cloud Application Platform (ICAP)</strong></p><p>ICAP is targeted at 6000+ WAS clients looking to modernize their applications. It helps the smooth transition from the WebSphere Application Server to the next generation IBM Cloud Private (ICP) cloud application platform. It includes ICP, WAS, Transformation Advisor and IBM Cloud Garage services to help accelerate the modernization of the applications.</p><p>Trasanformation Advisor is deployed on ICP, It introspects existing WebSphere deployments (it creates application inventory and assesses modernization effort, classifying it as simple, moderate or complex) and provides recommendations, guidance and artifacts for deployment in Liberty containers and Kubernetes clouds</p><p>There is another product that was featured in one of the sessions:<strong> Microclimate</strong>. It is a beautiful tool aimed at developers for the rapid onboarding and creation of apps based on microservices. It includes its own IDE, but one can bring his/her own (VS Code, Eclipse, Atom etc). Here is the link: <a href="https://microclimate-dev2ops.github.io/">https://microclimate-dev2ops.github.io/</a></p><p>Overall, I’ve heard several approaches from different parties how applications can be migrated:</p><ol><li>Example of a large UK bank:</li></ol><ul><li>Migrate the VMs to cloud</li><li>Modernize</li><li>Containerize</li><li>Convert into cloud-native</li></ul><p>2. Mainframe applications:</p><ul><li>Install ICP</li><li>On-premise decomposition of legacy applications into containers</li><li>Modernize in cloud-like manner</li><li>Extend them into public cloud</li></ul><p>3. Another approach:</p><ul><li>Use Transformation Advisor</li><li>Containerize them with Microclimate</li><li>Deploy the containers with UrbanCode Deploy</li><li>Monitor the deployments with UrbanCode Velocity</li></ul><h3>Developers</h3><p>One of the key sessions was devoted to application developers and open source. CEO Ginni Rometty lead the panel discussion with lead developers from several organizations. Developers and open source play a key role in the new Chapter Two phase of the scaling of digital and AI. As Bob Lord, Senior Vice President of IBM said: “Growth and innovation today starts with developers and data scientists”. IBM wants to help containerize mission-critical workloads, and bring them to the ML, and post-ML world where reasoning and cognition will play even larger role. That is why developers engagement is so important. This engagement requires a lot of ground level work where Red Hat was very successful at.</p><p>We’ve heard from the panelists that the developers want two things: to be on the right wave and monetization of their work. We’ll see how the things will develop in this new triangle between IBM and its clients with legacy applications, Red Hat with open source and developer’s communities, and the external market where AI is expected to generate enormous value.</p><p>Read the next and final <a href="https://medium.com/@sasha.lazarevic/think-2019-part-5-2de15b83adfe"><strong>Part 5</strong></a> of my Think 2019 report, with the review of IBM partnerships, education programs and my conclusion</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aQ453TOp98h9N7l2qcgtpg.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a2185c949fad" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Think 2019 — Part 3]]></title>
            <link>https://medium.com/@sasha.lazarevic/think-2019-part-3-4c8e8bba8008?source=rss-ad51a0e4e895------2</link>
            <guid isPermaLink="false">https://medium.com/p/4c8e8bba8008</guid>
            <category><![CDATA[think2019]]></category>
            <category><![CDATA[ibm]]></category>
            <category><![CDATA[ibm-summit]]></category>
            <category><![CDATA[quantum-computing]]></category>
            <category><![CDATA[power9]]></category>
            <dc:creator><![CDATA[Sasha Lazarevic]]></dc:creator>
            <pubDate>Tue, 26 Feb 2019 13:23:42 GMT</pubDate>
            <atom:updated>2019-02-26T17:22:31.359Z</atom:updated>
            <content:encoded><![CDATA[<h3>Think 2019 — Part 3</h3><p>In <a href="https://medium.com/@sasha.lazarevic/think-2019-part-2-b034ec86790a"><strong>Part Two</strong></a>, I wrote about IBM’s Cloud strategy. But the infrastructure is what powers cloud computing.</p><h3>Infrastructure</h3><p>This part was very rich with demonstrations</p><h3>Power9</h3><p>IBM showed new and improved Power S922 server. In the standard edition, it works in similar ways like previous Power machines. But as of now, it will also be available in IBM Cloud, not only on-premise. General availability is as of now in the US , and starting from Q3 also in Europe. The big news is that the customers who use Power on-premise in their own data centers, can also benefit from new consumption-based subscription model, similar like being in the public cloud. This is very interesting for companies that prefer a lower upfront payment and shorter time commitment. The prerequisite is Power S922 model with minimum 12 months commitment.</p><p>Power S922 is much more powerful when used with GPUs. It is actually the best machine in the market for deep learning workloads. It features NVLink between the CPUs and GPUs with 100GB/s bandwith for 6 GPUs and unified memory space for all CPUs and GPUs. Ideal use case is as an accelerator of deep learning algorithms. And together with Power AI (set of libraries for machine learning), Power S922 will be also available in IBM Cloud as <strong>Watson Machine Learning Accelerator</strong>. Watson ML users will be able to select it as one of the training environments. This is an amazing machine, which is also used as building block for IBM Summit , the fastest supercomputer in the world</p><p><strong>IBM Summit</strong> has been installed in Oak Ridge National Lab, and is the only machine in the world that can reach 1 Exaflop per second. It is consisted of 256 racks full of Power S922 with 200’000 cores and 27’000 GPUs. It is currently used for medicine, astronomy and climate change related research. It is a master piece of system design. You can see here how one rack looks like.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/768/0*GSDMQYzN5QkSHjON.jpg" /></figure><h3>Quantum</h3><p>Huge interest around the booth with quantum computer and for quantum presentations. Visitors could see super-conducting quantum system with the real chip and wiring for microwave controls. No new announcements for Think 2019, the new machine with 50 qubits is still cooking, nothing is known about its performance, the architecture or general availability. There was a number of noteworthy sessions around quantum computing and its impact on the future IT and various potential use cases.</p><p>But we’ve learned that IBM Research is very busy with working on new quantum-safe lattice-based encryption algorithms. It will not rely on the product of two numbers (like RSA), but on proximity function in n-dimensional space. First offering is expected for zSystems, followed by HW appliances. Here below is how it looked like. While waiting for new 50 qubits machine, practice on the smaller one : <a href="https://quantumexperience.ng.bluemix.net/qx">https://quantumexperience.ng.bluemix.net/qx</a> . You can also play a quantum video game on your smartphone : <a href="https://helloquantum.mybluemix.net/">https://helloquantum.mybluemix.net/</a> . Quantum is fun, and you will learn a lot about quantum physics (even more than you ever planned</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*nQrVanSisFPt-ObM.jpg" /></figure><h3>Mainframe</h3><p>Mainframe is continuously being updated and prepared for the hybrid cloud. IBM showcased several new features :</p><ul><li>zDevOps with UrbanCode Deploy for z Systems</li><li>Containers for z System through ICP for LinuxONE</li><li>z/OS Cloud Broker for ICP, which exposes and integrates z/OS based services into ICP for modern, container-based application development.</li></ul><p>Big idea is actually that external, open-source developers’ communities will have better and easier access to deep IBM resources when modernizing these mission critical applications. Don’t forget that 90% of big banking transactional systems run on IBM mainframes.</p><h3>Storage</h3><p>No big news here. IBM showcased some of the existing technologies like FlashSystem 9100, Storewize V7000, and IBM Spectrum Scale which is used in IBM Summit supercomputer. There is a new and interesting product <strong>Spectrum Discoverer</strong> that creates metadata of the existing data, and emphasis on hwo IBM performs HW-based storage encryption of the files that need to be stored off-chain for Blockchain applications.</p><h3>Databases</h3><p>A lot of talk and some interesting announcements :</p><ul><li><strong>Db2 AI database</strong>. This is very powerful database for ML. It includes data virtualization so that no ETL is needed, open-source drivers for Python, Jupyter Notebooks, Node-Red, and it comes now with self-tuning and ML-based SQL optimizer. The objective is to accelerate AI development.</li><li><strong>Support for multi-cloud</strong>. Db2 Datawarehouse is also available on AWS</li><li>This is part of larger portfolio named <strong>IBM Hybrid Data Management Platform</strong>, which includes Db2 Database, Db2 Datawarehouse, and can be consumed as service either on Cloud or on-prem</li></ul><p>All in all, the infrastructure, whether standard, high-end, or extreme gives IBM absolute advantage when it is about private portion of the hybrid cloud, in terms of performance, reliability, scalability and flexibility.</p><p>You can continue reading my story about Think 2019 in <a href="https://medium.com/@sasha.lazarevic/think-2019-part-4-a2185c949fad"><strong>Part 4</strong></a> of this blog post about IBM news in the domains of Blockchain, Microservices, and Application Modernization.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aQ453TOp98h9N7l2qcgtpg.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4c8e8bba8008" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Think 2019 — Part 2]]></title>
            <link>https://medium.com/@sasha.lazarevic/think-2019-part-2-b034ec86790a?source=rss-ad51a0e4e895------2</link>
            <guid isPermaLink="false">https://medium.com/p/b034ec86790a</guid>
            <category><![CDATA[think2019]]></category>
            <category><![CDATA[ibm]]></category>
            <category><![CDATA[cloud-computing]]></category>
            <dc:creator><![CDATA[Sasha Lazarevic]]></dc:creator>
            <pubDate>Tue, 26 Feb 2019 13:17:07 GMT</pubDate>
            <atom:updated>2019-02-27T17:16:47.697Z</atom:updated>
            <content:encoded><![CDATA[<h3>Think 2019 — Part 2</h3><p>Here is the continuation of my summary of Think 2019, the main IBM conference in 2019. In <a href="https://medium.com/@sasha.lazarevic/think-2019-part-1-e2c4223944fb"><strong>Part One</strong></a>, I write about IBM AI and Data products, and here I go into details about Cloud.</p><h3>Cloud</h3><p>Everbody knows that AWS dominates in the cloud industry. Many serious players like HP, VMWare, Cisco and others have already left the market. There are still companies that are growing like Microsoft and Alibaba (with Aliyun about which I wrote already back in 2014). IBM is present in this market since the acquisition of Softlayer in 2013, with IaaS, PaaS and SaaS offerings. As I have already mentioned, Ginni Rometty stated that IBM is orienting itself towards the future where hybrid and multi-cloud will play even bigger role than it was the case until now. So, what are the main takeaways of Think 2019 regarding Cloud ?</p><p>First, I would highlight the joint keynote speech with Jum Whitehurst, CEO of RedHat. Several months ago, IBM and Red Hat have agreed the integration where IBM is buying Red Hat for $34B. Red Hat is bringing to IBM its portfolio of open source and cloud enabling tools and projects like RHEL, JBoss, Ansible, OpenShift, and many others. RedHat has very strong culture of open source development (which is not strange to IBM), and a large community of almost 8 million developers. IBM counts to be able to scale Red Hat business to a new level and provide this developer community the access to its deep industry expertise and large client base.</p><p>Next, showcasing<strong> IBM Cloud Private (ICP)</strong>, which is based on Kubernetes. ICP aims to bring the cloud discussion on a new level, a layer above traditional VMs and a step in the future towards application microservices. ICP runs on top of ESX, bare metal machines, Power, mainframe, OpenShift and is as of now multicloud ready. ICP is the tool that will allow IBM and Red Hat to perform on-premise decomposition of customers’ legacy applications into containers, so that IBM clients can first manage their applications in cloud-like manner, and then eventually extend them into public cloud. This now becomes possible with the help of Red Hat and their developers communities.</p><p>Further, <strong>IBM Multicloud Manager</strong>. Multicloud Manager will help to manage Kubernetes contaners over multiple cloud providers in a consistent way. When you think about multicloud management, and most organizations even now use several (on average 5 different cloud providers), several issues become very important and difficult to manage:</p><ul><li>data transfer</li><li>connectivity</li><li>consistent management of these disparate environment.</li></ul><p>Multicloud Manager helps to bridge these gaps and unify the whole environment. It is bundled with the new version ICP 2.1.3. When you install it on ICP, it becomes your hub cluster. It is enough to install agents (Klusterlets) on other ICP clusters (IBM Cloud, AWS, Azure etc) and they will become your managed clusters. You can then monitor the dashboard for the aggregated information across all your clusters, view the topology of resources within those clusters, or see more details by drilling down into individual clusters.</p><p>This tool is different from IBM Cloud Automation Manager, which is used for automation of IaaS (VMs) provisioning with Terraform templates and interface with Chef for configuration management. Multicloud Manager works with Kubernetes.</p><p>IBM has showcased another new platform meant for the multicloud managed services: <strong>Multicloud Management Platform (MCMP)</strong>. This platform supersedes and replaces earlier Cloud Brokerage and covers platforms like IBM Cloud, AWS, Azure, Google, zCloud, VMWare etc. Optionally it is used together with IBM Services Platform with Watson (ISPW) for intelligent and automated operations. MCMP can be deployed by the client or provided as managed service. It can be shared or dedicated. It leverages the new extended partnership with ServiceNow (as of November 2018) so ServiceNow is the central component with integrated self-service portal to configure, buy, and deploy services from AWS, Azure, IBM Cloud, VMWare, Google etc.</p><p>A host of other tools for the DevOps, Operations, and Governance, for multicloud, IaaS private cloud (VMs) and containers (based on ICP). I have seen well-known tools like APM for the application and infrastructure monitoring and management, but also ScienceLogic for multicloud management, as ScienceLogic has some very nice plugins for all these here-mentioned public cloud providers. Cloud provisioning can be done through many channels like vRealize, IBM Cloud Automation Manager, OpenShift, and Ansible. ICPW gives AIOps capabilities for system optimization recommendations and ML-based anomaly detection.</p><p>Managed services for the multicloud that leverage this platform are called IBM Services for Multicloud Management. More information here:</p><p><a href="https://www.ibm.com/services/cloud/multicloud/management">https://www.ibm.com/services/cloud/multicloud/management</a></p><p>Here below is chart describing the structure of MCMP. So, Multicloud Manager is tool that enables consistent management of multicloud Kubernetes environments, and MCMP is much larger. It includes container clusters, but has also governance, monitoring, compliance, backup, service management and everything else.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*cnzeHDbQvS_FNtQX.png" /></figure><h3>VMWare partnership</h3><p>Think 2019 showcased and emphasized very strong and visible partnership with VMWare. I mentioned that VMWare uses Watson technology for their customer experience improvement, but the partnership goes very far. IBM Cloud for VMWare Solutions (IC4V) is a new service, which actually means the IBM hosting services and full VMWare stack on top of IBM Cloud bare metal servers.</p><p>In this context, VMWare spoke about three step approach how the clients can move to IC4V:</p><ol><li>Use VMWare HCX to easily migrate VMs from customers’ own data center to IBM Cloud. The existing customer’s network is extended to the IBM Cloud</li><li>Once migrated to IBM Cloud, augment the existing applications with other complementary IBM Cloud services</li><li>Install and leverage ICP / Kubernetes. Keep your databases on traditional VMs, and use containers for web servers and other application modules</li></ol><p>VMWare showcased <strong>NSX-T</strong> : This is a new protocol for network micro-segmentation that supports containers, VMs and bare metals. NSX-V is for VMWare only, while NSX-T supports VMWare and other hypervisors and multicloud environments, and it enables unified security policy model. NSX-V and NSX-T are not compatible, as the underlying encapsulation is not the same (NSX-V uses VXLANs, and NSX-T is based on new protocol Geneve). You can check <a href="https://console.bluemix.net/docs/services/vmwaresolutions/archiref/vcsnsxt/vcsnsxt-techpreview.html#technology-preview-of-vmware-nsx-t-on-vmware-vcenter-server-on-ibm-cloud">here</a> for more information.</p><p><strong>Veeam</strong> : Popular third-party solution for VMWare backup and restore. Veeam is IBM’s partner and it was announced during the conference that the new version of Veeam will have better integration with IBM Cloud Object Storage (ICOS), sot that unused backups can be offloaded to cheaper ICOS. This new enhanced version of Veeam also supports backups of SAP HANA and integrates nicely with Oracle RMAN.</p><h3>Cloud Security</h3><p>Cloud Security was a big topic at Think 2019. Here are some new products:</p><p><strong>IBM Cloud Key Protect</strong>: It helps provision and manage encryption keys for IBM cloud services and your cloud applications. The keys are secured by hardware security modules (HSM) based on FIPS 140–2 Level 2 encryption. There is no way the keys can be stolen. Even if the IBM cloud system admins try to intercept them or access them, the system stops any access to the keys or the data. This is suitable for Bring-Your-Own-Key (BYOK). Key Protect integrates also with IC4V for vSAN and vSphere encryption</p><p><strong>IBM Cloud Hyper Protect Crypto Services</strong>: This is your own dedicated HSM encryption management solution for Keep-Your-Own-Key (KYOK) use cases. It is single-tenant</p><p><strong>IBM Cloud Data Shield</strong>: This is solution for application developers, which allows to create secure enclaves where sensitive data can be visible only to the application and not to the underlying OS. Enclave Manager is also available as management console with reporting functionalities.</p><p>Overall, Hybrid Cloud and Multicloud were the key topics of discussion during Think 2019. IBM’s strategy that involves Red Hat in a very powerful way is all around hybrid cloud. But another key component enabling this strategy is the infrastructure.</p><p>To continue reading about the infrastructure (including IBM Summit, quantum, mainframe and the best machine for deep learning), go to <a href="https://medium.com/@sasha.lazarevic/think-2019-part-3-4c8e8bba8008"><strong>Part 3</strong></a> of my story.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aQ453TOp98h9N7l2qcgtpg.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b034ec86790a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Think 2019 — Part 1]]></title>
            <link>https://medium.com/@sasha.lazarevic/think-2019-part-1-e2c4223944fb?source=rss-ad51a0e4e895------2</link>
            <guid isPermaLink="false">https://medium.com/p/e2c4223944fb</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[ibm]]></category>
            <category><![CDATA[think2019]]></category>
            <category><![CDATA[project-debater]]></category>
            <category><![CDATA[watson]]></category>
            <dc:creator><![CDATA[Sasha Lazarevic]]></dc:creator>
            <pubDate>Tue, 26 Feb 2019 13:11:48 GMT</pubDate>
            <atom:updated>2019-02-27T17:16:14.313Z</atom:updated>
            <content:encoded><![CDATA[<h3>Think 2019 — Part 1</h3><p>Written by Sasha Lazarevic on February 26, 2019</p><p>Please note: I work for IBM, but all what I’m writing in this post represents only my personal opinion and is not an official statement of IBM or any of its entities. If you consider buying any of the IBM products, please consult with our sales representatives.</p><h3>Think 2019</h3><p>February 12–15 was the week for Think 2019, according to some analysts the most important tech event of the year. IBM assembled together all major individual exhibitions and this was the second edition of Think, now the main annual IBM conference. This time the event took place in San Francisco. I attended the event as one of the speakers on the subject of Watson Machine Learning and Watson Studio, and in my session I also had as guest <a href="https://anyline.com/">Anyline</a>, AI company from Austria.</p><p>The conference was successful and it was huge. Not only based on the number of participants (around 30’000), or sessions (more than 3’000), but also on the quality of discussions and topics. IBM events are not so much about announcements, but I will list and describe the most important ones in this post, along with highlights of the most recent and interesting technologies.</p><p>But the most important were the speeches and discussions around the IBM strategy and intentions. CEO Ginni Rometti opened the conference with statements that this new Chapter Two (which starts now) in the digital transformation is all about scaling of digital and AI, and moving mission critical workloads of large organizations to the cloud. She said that the future is multicloud, that many workloads will never move to public cloud and that in less regulated industries the ratio (public/ private cloud) will be 60:40, and in more regulated 40:60. What I have seen later, made me confident that IBM is well prepared for this Chapter Two.</p><h3>AI and Data</h3><p>Let me start with what was the dominant subject of Think 2019: AI and Data. The breaking news from the conference was Watson Anywhere announcement. Watson AI platform is going to be available on AWS, Azure, Google clouds and on-premise. So, those users who already have their applications and data with these cloud providers will be able to benefit from the full portfolio of Watson tools. I remember several months ago the initiative “Bring AI to the Data” and this is just a logical extension of that approach. I think this is excellent news for our customers and for IBM.</p><p>Next, four flagship AI products received very high visibility during the conference:</p><ul><li>Watson Assistant</li><li>Watson Studio</li><li>ICP4Data, and</li><li>Watson OpenScale</li></ul><p><strong>Watson Assistant</strong> is already very well-known tool , used by a large number of companies to improve their customer experience through modern and sophisticated conversation interface, together with other NLP and speech-related services. There are some new features that were already announced several months ago. What was showcased during Think 2019 are numerous examples of how different organizations from various industries implement and benefit from this tool. We have heard also that VMWare deployed Watson Assistant, together with NLC and NLU, to help their support professionals, and improve satisfaction of their hundreds of thousands of customers.</p><p><strong>Watson Studio</strong> is integrated development platform for Data Scientists and ML developers that has received a lot of attention and visibility during the last 12 months. I remember of a recent Gartner report featuring Watson Studio as one of the most modern data science platforms. New version includes very nice visualizations tools based on Cognos, and features like NeuNets for neural network synthesis and AutoAI for automatic ML feature generation, algorithm selection, and view of the progress tree through hyperparameter optimization. There is also integration with Watson OpenScale,</p><p>Watson Studio is now available on Cloud, Local (on-prem) and in Desktop version (on PC)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*4aOPSrw-hCbcTzp5.png" /></figure><p><strong>ICP4Data</strong> platform is equivalent to Watson Studio, but it is installed on premise (you need a 16 core server). It has even more capabilities for data transformation, including for example data virtualization, so for some use cases, you don’t need to move the data to ICP4Data. The data can stay where it is, on remote systems. You get better performance this way for example if you will select only small portion of data for your analytics algorithms than if you move it over. There was a lot of interest in ICP4Data during the conference. This couple of tools : Watson Studio on Cloud and ICP4Data on-premise are actually the cornerstone of IBM AI and Data strategy. They allow the data to be processed where it is, and AI models to be trained in the proximity of data, and everything to be managed and deployed in a consistent manner.</p><p><strong>Watson OpenScale</strong> is a new tool, just several months old. It has been created to address the need for trust and explainability of AI, especially of those models based on deep neural networks where machines learn themselves and it is very difficult to identify the logic of actually very accurate predictions. Bias in AI can come from various sources, either from the data itself, or from flawed models, or from the new data as it is coming during the production phase. OpenScale will monitor these models for bias while they are in production. When you start using it, you define the protected classes (gender, race, age etc). OpenScale will then perform what is called perturbation testing: it will withhold some data, and test to see if the model will show any abnormal predictions (gender predictions for example should always be 50–50%). If their are any deviations in this result, the model will alert for potential bias. OpenScale can identify bias even if some other features (like for example film preferences) can cause the model to generate gender bias. Very powerful tool.</p><p>The bottom line is that many organizations will create their own ML models, or download open-source models from internet to reuse them, and will eventually end up with many different non standard models. Management of these models then becomes very difficult, and that is where Watson OpenScale comes into play.</p><p>IBM AI portfolio is much deeper of course (don’t forget Discoery, Visual Recognition, Speech-to-Text and Text-to-Speech, and numerous other vertical applications for health, compliance etc , but I selected these four tools based on their transformation potential in the context of new IBM’s strategy for Chapter Two.</p><p>There were also some other very interesting news and demonstrations:</p><p><strong>AI Fairness 360</strong> : This is another bias detection mechanism that can be embedded into your own ML algorithms. It includes 30 different bias metrics. You can check here for more information on how it works and how to implement it:</p><p><a href="https://developer.ibm.com/open/projects/ai-fairness-360/">https://developer.ibm.com/open/projects/ai-fairness-360/</a></p><p><a href="https://www.youtube.com/watch?v=X1NsrcaRQTE">https://www.youtube.com/watch?v=X1NsrcaRQTE</a></p><p><strong>Adversarial Robustness Toolkit</strong> : This is excellent tool for the security hardening of your ML models. It applies different attacks and helps you to improve your model. You can check here on how it works and download the libraries:</p><p><a href="https://github.com/IBM/adversarial-robustness-toolbox">https://github.com/IBM/adversarial-robustness-toolbox</a></p><h3>Project Debater</h3><p>This was a great show. IBM has showcased AI solution that can debate with humans on various topics and make arguments to persuade in a discussion. The show was actually a competition between Project Debater and a human, Harish Natarajan, the world champion in political and social debates. The debate was organized around a topic: “Should preschools be subsidized by government ?”. Project Debater argued for and Harish against. The audience voted for or against this issue before the debate and voted again after the debate, and the winner was one who managed to persuade more people to change their mind to his/ her side.</p><p>Project Debater lost. But capabilities that this machine demonstrated were amazing. There were three rounds in which each opponent was able to bring his/her points and refute the points of the other side. Project Debater started the first. All sentences were completely correct and relevant for the topic, the structure of the speech was natural and effective, the argument was exposed in a very rational, comprehensive and structured way. And not only that. After exposing why, Project Debator started bringing dozens of relevant scientific studies, statistics, research reports, hard data about this subject. Multitude of data to support the decision. It was simply amazing. And simply impossible for a human to argue with such a machine which knows everything that can be known on this subject.</p><p>The second and third rounds were less successful for Project Debater. Harish didn’t know so many statistical and scientific data about this topic, but his way of arguing was, well, like we humans do: using many contrasting parts of the speech, finding holes and exceptional cases. He began the speech with “Yes, I agree with one point, but I disagree about the others” and used similar rhetorical figures. Project Debater was not able to follow these subtle nuances in the human speech where people expand the subject, then change the subject, and make it looks like your arguments suddenly become irrelevant.</p><p>I don’t know what algorithms are behind Project Debater, but it looks like there is a lack of a sophisticated attention mechanism that could identify all these rhetorical constructions and prepare the right arguing strategies. And to argue with humans, you need very sophisticated algorithms. We are getting there, but it takes time.</p><p>But even though Project Debater didn’t win, it surprised everyone. It was an amazing achievement. Competitors had 4 minutes to prepare themselves for the next round. In these 4 minutes, Project Debator had to use speech-to-text to create transcription of what Harish said, analyze the text, find the new arguments and prepare the speech, and then synthesize the speech based on that text. And all that had to be delivered without any grammatical or semantic inconsistencies.</p><p>What are the potential use cases for this AI debating engine ?</p><p>Some analysts say : marketing and sales. It can sell products or services better than humans. Also other professional and consulting domains, as it can help justify the decision based on hard data and scientific results. If you have one hour of time and you like to watch , check <a href="https://www.youtube.com/watch?v=6_J0iW1Ru7U">here</a></p><p>But maybe we should be careful to whom we give such a tool, as it is really powerful.</p><p>As a conclusion about this section, AI was central topic during the Think 2019 conference. But it is becoming the central topic for most of IBM’s customers. This is actually where we want to deliver the most business value. And to do that, IBM needs first to resolve the biggest problem of this Chapter Two — migration of mission critical workloads to Cloud.</p><p>Continue to <a href="https://medium.com/@sasha.lazarevic/think-2019-part-2-b034ec86790a"><strong>Part 2</strong></a> of this story to understand IBM’s Cloud strategy.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aQ453TOp98h9N7l2qcgtpg.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e2c4223944fb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Exploring Deep Learning and Neural Network Modeler with Watson Studio]]></title>
            <link>https://medium.com/ibm-watson/exploring-deep-learning-and-neural-network-modeler-with-watson-studio-ead35d21a438?source=rss-ad51a0e4e895------2</link>
            <guid isPermaLink="false">https://medium.com/p/ead35d21a438</guid>
            <category><![CDATA[watson]]></category>
            <category><![CDATA[neural-networks]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[tutorial]]></category>
            <category><![CDATA[data-science]]></category>
            <dc:creator><![CDATA[Sasha Lazarevic]]></dc:creator>
            <pubDate>Wed, 23 May 2018 19:43:03 GMT</pubDate>
            <atom:updated>2018-05-23T19:43:03.495Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/640/1*NSi_6E6RaNodwv268g0EvQ.png" /><figcaption>The Neural Network we’ll build</figcaption></figure><p>It was too much fun creating my own convolutional neural network with <a href="https://dataplatform.ibm.com/">Watson Studio</a>! <a href="https://www.youtube.com/watch?v=xLcHbF8GM9c">The Neural Network Modeler</a> provides expressive and intuitive graphical tools for building powerful deep learning models. It integrates well with all aspects of <a href="https://datascience.ibm.com/docs/content/analyze-data/ml-overview.html">Watson Machine Learning</a>. Application developers, analysts, and data scientists use WML to integrate machine learning into their workflow seamlessly. For example, you can save and deploy models as a REST API with a few clicks, minimizing the time it takes to put machine learning models to work. In this example, I used the familiar <a href="http://yann.lecun.com/exdb/mnist/">MNIST data set</a> (images of hand-written digits) and created a three layers deep CNN. Then, I exported the algorithm in TensorFlow / Keras python code before training my model using NVIDAI GPUs. I also used the downloaded code to train my model locally in my Jupyter Macbook environment. In both cases, I got some very nice results. Of course, training with GPUs in the cloud is lightning fast.</p><p>The images of digits we’re classifying look like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/594/1*Ft2rLuO82eItlvJn5HOi9A.png" /><figcaption><a href="https://commons.wikimedia.org/w/index.php?curid=64810040">https://commons.wikimedia.org/w/index.php?curid=64810040</a></figcaption></figure><p>To data scientists and AI developers : go and use it. And for all other enthusiasts, the best way to learn AI is to create it yourself 🙂 !</p><p>First, to access the Watson Studio environment, you need your IBM ID. Visit <a href="https://console.bluemix.net/">IBM Cloud</a> and then create your free account. <strong>Everything I’ve used here is free</strong>, so you can learn, develop your models and deploy them for productive use quickly and painlessly.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/972/1*FYGeqC5Sq8j2Ob7a0M5Eow.png" /><figcaption>Your IBM Cloud account</figcaption></figure><p>Once you have the access to IBM Cloud, you will need to source three services:<strong> </strong><a href="https://console.bluemix.net/catalog/services/cloud-object-storage"><strong>Cloud Object Storage</strong></a> (to store your data sets, trained models and training results), <a href="https://console.bluemix.net/catalog/services/machine-learning"><strong>Machine Learning</strong></a> (to train your models and benefit from ultrafast GPUs) and <a href="https://console.bluemix.net/catalog/services/watson-studio"><strong>Watson Studio</strong></a> (to create and manage your DL models, collaborate with your team, and more).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WTJ2ODgRhuC1FbTtmn4vKA.jpeg" /><figcaption>The services we need</figcaption></figure><p>When this is done, check in your Watson Dashboard that you have all these three services properly created. You can check this from <a href="https://dataplatform.ibm.com/console/overview"><strong>within the Watson Studio Admin Console</strong></a><strong> </strong>or from <a href="https://console.bluemix.net/dashboard/apps"><strong>within your IBM Cloud Dashboard</strong></a>.</p><p>The next step is to <strong>create buckets in your Cloud Object Storage (COS)</strong> and <strong>upload your data</strong>. My recommendation is to <strong>create two buckets</strong> (you can also call them containers for your data objects), <strong>one for the data</strong> and <strong>one for the models and results</strong>. For larger models, more granularity is needed, since the Watson Machine Learning (WML) will load the data from the training data bucket, so you want to avoid unnecessary waiting time and memory consumption for the data which is not used in your training runs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cwbwXWED6IY9KKYh73HOkw.jpeg" /><figcaption>Creating training and testing buckets</figcaption></figure><p>Here is how it looks like when you create both buckets and upload the data:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gr2ANg2IwVkkWnBNhOIbLQ.jpeg" /><figcaption>Notice the two buckets</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UrAJ6MTNjGxUaB2lBTEuAA.jpeg" /><figcaption>We’ve added the data to our training data bucket</figcaption></figure><p>When this is done, you can launch Watson Studio and create your project. I named mine MNIST-LZRVC. You will see on the right hand side your Cloud Object Storage (COS). Watson Studio will automatically create connection to your COS service.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iwZXaolo7hC_s-O2VkSsqw.png" /><figcaption>New project home screen</figcaption></figure><p>Next, in the assets tab of Watson Studio, go to the Modeler Flow and add a new flow. Name it like your project but with some unique ID so that you can recognize when you start modifying and experimenting with different architectures.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/756/1*4--fDCMJJAyMRBWFz672lA.png" /><figcaption>A new Modeler canvas</figcaption></figure><p>When you do this, you will land on the design canvas for modeling your neural networks. You can click on the left upper button to open the side toolbar with different groups of NN building blocks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XORNr7iIZBrol_h2H2fQeQ.png" /><figcaption>Expand the side bar</figcaption></figure><p>These building blocks allow you to construct the whole deep neural structure from the scratch quickly and easily. There are also many samples available in the platform. Of course, you need to know what you are doing, but there are quite a lot of very good learning resources on the web if you want to learn more. In my particular case of this image recognition example, I have connected my input data with three convolutional + ReLU activation + pooling layers, then to a fully connected layer and passed the resulting feature representation to softmax classifier with 10 categories (for 10 different digits). In the end, I connected everything with the cross-entropy loss function and used Adam adaptive learning-rate optimizer. I’m measuring the performance of my model on the validation data set using the accuracy metric.</p><p>I have experimented with different number of layers. One is clearly not enough, but three layers give very good results, even after only two or three training epochs. Of course, you need to be careful with the hyperparameters like number and sizes of filters, strides, channels, etc.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/880/1*q9KwAnEAWnZBb6quh9r1sw.png" /><figcaption>Architecting the network</figcaption></figure><p>When you have done this the most creative part of your project, connect your model with the COS bucket that contains your data set. Click on the three dots of upload your files in this new data bucket. I’m using pickled data in three separate files for training, validation and test.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UAnNufVrh_FessQ3FqIcOA.jpeg" /><figcaption>Be sure to configure the Image data node</figcaption></figure><p>When your model is ready, take a look at the two arrow-like icons in the toolbar. They allow you to download the code or launch your model.</p><p>I will first download this model on my laptop, because I want to play with it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CAF7XbPdDiTdPFsHLUZvDw.jpeg" /><figcaption>Download the code</figcaption></figure><p>When you select this download button you will get the choice of different formats. Pay attention — with Watson Studio y<strong>ou can create training code for multiple deep learning frameworks</strong>. This is the beauty — using IBM’s Fabric for Deep Leaarning (FfDL), you are able to run your deep learning algorithms on TensorFlow (with or without Keras), PyTorh and Caffe, all that within one unified training and deployment cloud environment.</p><p>So, I continue to download my model. I’m interested to see how it is written.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*IfKPoAXD24lI_nLc.jpeg" /></figure><p>And here it is, I’m running it locally in my laptop python. No changes to the code needed !</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/809/1*VOVKj8WD6SAxRE3HvdrGvA.jpeg" /></figure><p>I also like very much Jupyter notebook environment, so I just copy-pasted my code into it. If you would like to check it out, please download all files from my github repository.</p><p><a href="https://github.com/LZRVC/mnist-with-keras-and-watson-studio"><strong>https://github.com/LZRVC/mnist-with-keras-and-watson-studio</strong></a></p><p>I’m adding here a couple of screenshots to show the training and prediction with this Jupyter notebook. One training epoch takes 2 minutes, and I’m getting 96% of accuracy on the included test data set !</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*n9QKDbi-GbR_yffnuqgBYQ.jpeg" /></figure><p>Here you can see the structure of the model :</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/698/1*cbRJJtV_uKh7smx4lUE6vw.jpeg" /></figure><p>And here is the prediction part. For the previously unseen images, I’m opening them in grayscale, adding the contrast, normalizing, resizing and reshaping them. Then, I use <em>model.predict()</em> method on this image and extract the predictions from the classifier. This is an array of 10 probabilities, where in the end we select the highest probability for the image to deliver our prediction.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/857/1*PAucW5jdUYv8VVgO3eEIuw.jpeg" /><figcaption>Working with my local Jupyter</figcaption></figure><p>Note : I noticed inaccurate predictions for number 7, when I write it with a dash across the numeral, and for all numbers when they are written with thin pencil. The training data seems to be of a similar type, so the real-world results are a bit skewed.</p><p>Now, going back to Watson Studio. I want to upload my model directly from the design canvas to Watson Machine Learning engine.</p><p>First, we will need to associate WML to our project. Select your project, and Setting menu tab option. Scroll down to see the Associated services, and add Machine Learning</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QGv8bwAd6Wuc_c5tvww29w.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/559/1*3AulHsmIdPOfWeJlfm_zYA.jpeg" /></figure><p>Now, go to your neural network modeler part of Watson Studio and click on the other icon on the toolbar to publish your model to WML.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/759/1*fUfY6L25UvEyLxkskjpFig.jpeg" /></figure><p>We will now create an experiment (in other words, prepare to launch the training run):</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yElK68ICFek6gWnONwT9kg.jpeg" /></figure><p>Enter a name of your training run (mind using IDs and some meaningful abbreviations so that you can trace back where the results are coming from). Select and confirm your COS buckets for the input data and results, and create your training definition to choose the number of GPUs and if you would like to do hyperparameter optimization (Select “None” — I will describe this in one of the next posts in the future).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wUHFwO8EFa_aF2TmqyGn0Q.jpeg" /></figure><p>When you click on Create and Run, WML will automatically start your training process. You can monitor the progress, review the log files as the system works. There is some delay until your job lands on the GPUs, and here after 5 minutes 25 seconds I got the results. This elapsed time is actually not bad, having in mind that I left 10 epochs specified in my model definition. One training epoch took 18 seconds, and there was also some queuing of my job before it started.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0vA7Jg3E4x71r9yE_2kvVw.jpeg" /></figure><p>Here in the results page you can see the location of your trained model. In the next blog post I will introduce more sophisticated concepts showing WML can be used for your Deep Learning projects, including Hyperparameter Optimization.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9fYylHQAGwlcGKTzMKpbHg.jpeg" /></figure><h3>Conclusion</h3><p>Watson Studio, Watson Machine Learning, and Cloud Object Storage are, as you could see from this tutorial, very sophisticated, well integrated, extremely flexible and easy-to-use tools for data scientists and AI developers. DLaaS provides excellent features that make IBM Watson Studio second to none. <a href="https://www.ibm.com/cloud/watson-studio">Get started for free today!</a></p><p>Sasha Lazarevic, IBM<br>May 21, 2018</p><p><a href="https://www.linkedin.com/in/lzrvc/">https://www.linkedin.com/in/lzrvc/</a><br><a href="https://github.com/LZRVC/mnist-with-keras-and-watson-studio">https://github.com/LZRVC/mnist-with-keras-and-watson-studio</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ead35d21a438" width="1" height="1" alt=""><hr><p><a href="https://medium.com/ibm-watson/exploring-deep-learning-and-neural-network-modeler-with-watson-studio-ead35d21a438">Exploring Deep Learning and Neural Network Modeler with Watson Studio</a> was originally published in <a href="https://medium.com/ibm-watson">IBM watsonx Assistant</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>