 Yn arweinydd hwnnw. Fy oedd y ddechrau. Mae'r ffordd yn ddalamol yma yn ymgyrch yn gyflawni Jayce. Fy oedd yw'r cyflwyll yma'r ddalaes iddyn nhw'n gyflawni'n gyflawni, yn gyfer y gyflwyll yn gyflawni. ICHEC wedyn yn gael ydyn nhw'n gweithio'r ddalaes a Jayce erbyn i'r cyflwyll yn gweithio'r ddalaes. Mae'n gweithio'r cyflwyll ar gyfer y gyflwyll, yw'r rhaid iddo'n gweld eich ffobl o'r wider ac oedd yn y dweud o'r bobl yng Nghymru, ac mae'n hynny wedi'i gwelio gwahanol o unrhyw unig o'r Unedd, yn ffawr, yn chefiwyr, yn ysgledig, yn unig o'r unedd. Ydyn nhw'n byw'n cael ei wneud i'r ddisgwst, os yw'n cael ei rhai, mae'r cychwynod o bobl yw cyfnodol, cychwynodol a'r cyfnodol. Rydw i'n edrych, mae'n ddysgol y oedd ychydig o'r law fforddau Mor's sefydlu gwylliant o'r bwysigol yn cyfnodydd cyfnodol, o'r bwysigol, mae'r bwysigol yn cyfnodydd. Ond mae'r bwysigol yn cyfnodydd. Rydw i'n edrych ar y cyfnodol yn ymwneud o'r ysgolionwyr ar gyfer o'r ysgolionwyr. Felly, mae'n ddweud i'r ffawr o'r cyfnodau ar gyfer y mylion, yn ymddangos, byw, erio, nautic, yn cynedd, ond yn fawr, mae'n fawr i'n ddweud, fawr i'n ddweud y cwestiynau. Mae'n ddweud. Ddwy'n ddweud i'r ffawr i'r gweithio. Ddwy'n ddweud i'r ffawr i'r gweithio. Mae'r cerdyn sydd yn ei wneud i'r hyn y gwaith ymrwy eto yn ystod y byddwrn ar y ddweud cyd-dweithio'i byw, y dyfodol cyd-dweithio ar yr ysgol, ond rwy'n fawr o'r holl honno'r gweithio'r roedd ymell. ac y ffynol yfodol o'r ffynol sydd wedi'u cyfnod, ydych chi'n gyfnod o ddweud o'r pwylltau a'u ddweud o'r widio a'u ddweud o'r ddweud. A gyd, mae'n gwneud o'r ffynol yn dweud o'r cyflwyffyr wedi'n nesaf o gydag a'u rydyn ni'n meddwl. Felly yna yw'r cyffredinol yn y cydwysaeth o'r ddweud o'r awdurdod a yna rydyn ni'n fag ar y gallwn mynd i ddweud, Ac rwy'n eu cyflawn i'r pethau sydd yn fwy o'r wych yn bwysig o'r gydafeydd. Rhywbeth mwy o'r ymddorol, yma'r chynyddiadau cyd-dwynger iawn, oedden nhw, yn cyflawn llwyngol ac yn ymddorol. Mae gyflawn i'r cyd-dwynger i'r cyd-dwynger o'r cyd-dwynger. Mae gennymau yn y ddechrau sy'n gweithio cymhwysig gyfmau Cymysgar yn gyfnodol, a byddai'r desgwyr, cyfan, ystafell, ac ymnodol. Y gwirionedd y cadw rhan, fy oedd 59% i 65%, ac nid os sy'n ni'n ei byw, rhaid i'r perlwysio gwirionedd y môl. Cymysgu'n rhaid i'r perlwysio cyllid yw 95% i 75% i Tsigwyr Elidwyd. Mae'r gwrthon o'r gwrthon yn ystod ardweud yn ddweud o'r llwgr yng Nghymru, yn olygu'r cyfnod o'r cyfnod. Mae'n ddechrau y bryd i'w ddod yn ddiddordeb yn ddiddordeb sylwf yn ddiddordeb sylw'r cyfnod, ond mae'n ddiddordeb yn 2005. Mae'n ddiddordeb sylwf yn ddiddordeb sylwf yn 1965, yn ddiddordeb sylwf yn ddiddordeb sylwf yn 2005, ond mae'r gafodon yn ddiddordeb sylwf yn ddiddordeb sylwf yn 2015. Ac rwy'n dechrau byddai y progres, Ac mae'n mynd i'n gwybod ddau'r cyfwyr yng Nghymru o'r cyfwyr yn ddiweddol, o'r cyfwyr gwyrdd cyfwyr. Rwy'n gwybod i'n gwybod, mae'n rhoi ddiwylliant yn ddau'r cyfwyr. Mae'n meddwl gydig o gydweithio'r cyfwyr o'r cyfwyr yn y cyfwyr, oherwydd, dyma'r cyfwyr yn ddiweddol. A'r cyfwyr yn ddiweddol, mae'n ddiweddol o'r cyfwyr yng nghymru I ddaf glau'r rydw i'w gweithio y llwyddoedd yn ddianfodol. Mertho gwrthio'r etymlu oherwydd mae oherwydd fynd i ddweud o gyfarion a fyddai. If mae'r ddweud o g挖fynn Fleiaf, roedd o gyfarion, yn 1978, iawn yn gweithio yn fynd ar Fflaid Llywodraeth ar New York yn y ddweud o gweithre'r llwyddoedd ac mae'r gallu mwynhau yn ddweud o'r ddweud o'r ddweud o gyfarion ar gweithre'r Llywodraeth. Ond oedd gofynnodd yn phoedd yn gallu meddwl i'r werthion, Gwsdalol ffordd o'r byw y gwisw'r wrth i gydhanol ar gyfer y dyliarol a i'r dyliarol, ynween amlŷ i ddod o'r cyfrifwys ac mae'n rhan o'r prosesoedd yn ei fitd. Ond o'r gwybodaeth a'r gwybodaeth yn y cyfreonol sydd ynnu'n olygu dda, ond unrhyw o'r argyfnod o'i ddweud â'r G. Yn y roedd y gynhyrch yn y myfwile i'r wneud digwyddag yn ôl? Yr argyfnod o'r gwybodaeth yn ôl? Y ddiwedd, mae'n meddwl i'r Morzlof yn ôl, mae'n ffrifadau, mae'n meddwl i'r Morzlof yn ddiwedd, mae'n meddwl i'r Morzlof yn ystafell ffordd, a'r ddiwedd yn meddwl i'r ddechrau, a dylai'r ddysgu am rhan oed yn y ddechrau. Y ddau mewn 60 yma, mae'n meddwl i'r byddwch ar y ger, i'n meddwl i'r byddwch yn y 70, mae'n meddwl i'r byddwch yn bwysig, mae'n meddwl i'r byddwch ar y byddwch ar y byddwch. this is what we call the top 500, in the early 90's a group of computer scientists, American computers scientists said, we want to have a registry of all the most powerful supercomputers in the world, because all domains, academia, industry, government, whatever you use it for. And we want to have a standardized benchmark, to see who is the billion one, that's a big concern for American Cuando. Felly, fy cheif wnaeth y posiedd argyferwad, sy'n fan mewn diemddugol, o'r number 1 cynsianwyr yw'r newydd o'r number 500. Mae'r number 500 Cadwyr yw'r number 500. Mae'r number 500 Cadwyr yw'r number 500 Cadwyr. Rydyn ni'n ddelig iawn o'r number 500. Felly mae'r number 500 Cadwyr yw'r number 500 Cadwyr. Ac yn iawn, mae'n amser, a dyna'r iawn yn yn gweithio yn iechyd a cael mawr fawr mewn. Mae'n adegwyd yn wedi'u'n mae'n cychwyn 15 oes gennymau ar yr iawn. Mae yna dwi'n dechrau i eich hollwyddiad cydwch. Os ydych chi'n amser, yna myllai Walton hereau. Walton isrwyd yn y Fathlwn Gwerthgawr, cws iawn i Llyfrgellol. Merthyn yma, y fathlwn gwirionedd thef i diodel at 400 yma, i gof ychydig i fyno eu gydweithio mewn Fathlwn Gwerthgawr, a mae'r pwysig o'r cymdeithas o'r gweithio cymryd yn cyfrannu. Yn y bwysig hwyl, y 7 ychydig, y cyfrannu yn ôl y 1. Y rhan o'r cwm yw'r pwysig yn ôl i'r uchydig o ffrindwyr o'r cofio'r cyllid ar gyfer hynny, felly mae'r pwysig yn olygu y 15,000 bwysig o'r cwm. Ychydig, o ble mae'r pwysig? Yn olygu, mae'n olygu o'r cwm yw 1,1 miliwn, is what I'm doing with technology in line with the accessibility that technology provides me. What's very important is one of the reasons why you are outperforming North Europe is that investments don't remain constant. There is growing global recognition of the economic and societal impact of high-performance computing across most governments in the world. Currently, in fact, there is a global race taking place between Asia, the USA and Europe. The mission of Europe currently is to become the number three. But China in particular and the US are firmly anchored in the doctor position so it might change with the current administration. There is a very interesting policy paper called HPC Europe's Place of Global Race that outlines the importance of high-performance computing for innovation and competitiveness in Europe. Now we were extremely fortunate for our 10th anniversary to have a visit from Dr. General Roberto Fiora who gave an address after the IAEA as it happens. We had to share Roberto between our two organizations and he provided a key note for half an hour to emphasise the importance of HPC, for example, to understand the function of the human brain, to model the human brain functions, in order to try to better understand the disease of ageing, for example, the kind of Alzheimer's disease and so on. I think you are similarly supportive and I read recently that the Vice President, Mr Hansi was also delivering a very important talk last week on the importance of high-performance computing. So if you wonder what the number one in the world looks like, this is what it looks like, it's in China, a couple of statistics which are interesting. First it's 10 million processing elements so if you want to use the world power of the world machine you need to understand how to make this 10 million so different little computers work together to solve the problem and this is part of the difficulty. But the second number which is very important is 15 megawatt, absolutely a lot of electricity, not only in terms of cost but in terms of infrastructure to actually support the world side. This is the last time we are going to see this problem. So something struck me with Moore's Law. Moore's Law in many respects I think under-represented the level of innovation that went into designing those chips. So if you're in your curve and you think, oh yeah, sure, just more of the same. They kept on doing the same thing, kept on going up. It's not at all the case. I mean our friends in the electrical engineering world and I should be careful, maybe they are electrical engineers today in the audience, but kept on seeing in the near term a few years ahead the limitation of the current process and they know that in order to keep on pressing ahead they will need to come with some form of innovation to overcome this barrier. So we went on an electrical engineer and those are the Intel guys kept on coming up with new processes to push the boundaries from micrometers to 100 nanometers to 10 nanometers and I believe next will be 7 nanometers which makes me wonder why it's going to stop. But the same is true for either one's computing. So it was linear. We went through a number of paradigm changes or a number of different approaches on how to solve a problem. One single is new machines, keep on going faster, it's magic. Vector machines, more for engineering, for weather forecasting, so all the big automotive industry and weather forecasting engines is one of those extravising machines. MPP stands for massively parallel. So we've got a lot of very small processing elements that we're using in the machine. Then we moved in 2005 the concept of mutico so most of you might have heard that even your phone or your laptop is dual core or quad core or eight core. Now we're pretty much at 20 cores plus. And now we are in the minicore era which is again thanks to NVIDIA, Intel and those are manufacturers. We have now little processors or cars you stick in your computer that have literally thousands of processing elements in your hand. So the latest Intel minicore proposition is called Intel Night Snonding offers you four times the power of the iPhone. So I like to compare it with the iPhone because it's interesting to put in perspective everybody has a mobile phone. Everybody can understand how much power actually this kind of device is offering. So don't be fooled by the linear curve where things are a lot more complex than they are. We have to face increased complexity in the architecture. I also wanted to put it into the cloud because people try to say the HPC guy hates the cloud guys and the cloud guys hate the HPC guys and it's not true. I really don't want to make a happy world. So it's also a sense of misunderstanding of these two technologies. The cloud is just for me a delivery mechanism and I have somebody from Google here so I need to be careful. But yes, it is a delivery mechanism and it is true to say that quite the contrary, the cloud is going to democratise the use of high performance computing by making it more readily available through public cloud services. So you look at Microsoft as you as one of the good examples of a cloud with good HPC resources available. But one thing is true though is that debugging fixing your code when it fails called debugging or optimizing your code making it go faster will be more difficult within the public cloud than it would be if you got a machine in your own data centre. So it was a call for anybody from the department of jobs if you have some different department of jobs or education or co-founders that don't be fooled HPC centres are going to be very quiet even more so in a cloud world. I'd like to say thank you to our friends from Accenture Labs. So one of the get meetings within the 10th anniversary we had the honours of getting something called Mark Calbillard who is a global managing director for innovation and he met his team and he totally inspired about quantum computing. What's next? What's after conventional computing? And they actually authorised me to use some of the materials I'm going to publish very shortly for current report on quantum computing. But I think you can put it in a really interesting context. We had a bit to walk down those blinker 1970s to whatever 1980s to 1115. We need to go back to the early days first computers to the next generation. What's coming next after conventional computing and this is quantum computing. Now what is quantum computer? Well first of all I think it's a piece of you appreciate the quality of the engineering. It's truly fascinating. This is the actual infrastructure that grows around the chip and this is part of the cryogenics part of the world cooling infrastructure. It's truly fascinating. This one comes from D-Wave I think we had one of the co-founders and chief scientific officer actually came to the visitors last year or two years ago as part of the bell days in November. It was a very interesting very interesting presentation. So a few points very quickly. Yes it is a genuinely destructive technology. This terminology in my view is used far too often incorrectly. Second point. It's not a panacea. Quantum computing will not replace mainstream computers. It's not to be used for general purpose. But it's particularly well suited to address any number of classes of problem optimization something at machine learning of the very broad range of the workforce. So it will coexist with conventional systems and again what I put here in double quotes are rotations from the actual report I mentioned earlier. As such it is unlikely classical computing will be replaced by mainstream by quantum computing. And our friends also forecast that consistent enterprise use is only 2 to 5 years away from one class of quantum computers called adiabatic. It is a category of systems the way are manufactured. 2 to 5 years from now, sorry. So I don't know about you but when autonomous vehicles started coming on the road I didn't realize I was already on the road in testament. When I'm being told quantum computers are already in use I thought it was just some kind of 10 years on the line I didn't realize initially that this technology was actually so close for being used for operational use. Absolutely, absolutely fascinating. And that demonstrated need to be extremely agile that in terms of police making when opportunities like this come you don't have the luxury to spend 2 years thinking about it because if you do, that cheapest side and all the countries will have made progress and you'll be being catch up. So if you don't want to be one of the innovation followers if you want to become part of the innovation I'm not sure whether it's actually setters or part of the innovator class you need to act quickly on this type of opportunity. And the final observation of how fascinating as well is when they say the number of qubits so the qubit is a small element in quantum computing if you look at a classical computer you've got a bit which is 0 or 1 so yes or no. In quantum computing it's called a qubit it can have several different properties that can influence each other anyway but it's notice that it's growing at similar pace as morthlaw and now the way that it sells starts taking order for 2,000 qubits computer. Morthlaw might seems to be relevant with classical computing but he may very well continue to be relevant within the quantum computing domain. So here I picked up a number of technologies I found particularly interesting and postulating that similarly the future has already started and again that sounds exhaustive some of you might disagree but I pointed out that deep learning of autonomous vehicles, blended reality you can break the slide. Now I like that quote about autonomous vehicles from Elon Musk a well-known entrepreneur and he described autonomous vehicles autonomous cars as a supercomputer with four wheels. It is not the exact quote but something of that nature that emphasised how important HPC was to this kind of technology and I think again it is something we miss far too often. When I mentioned high-performance computing how do you see people just switch off? I said I have a super computing thing no it's not. Transcomputing supercomputers are just the deep of the iceberg. HPC is a methodological approach to problem solving which is about exploiting the computing power of various components that will come back in a minute. I said what all of this technology 9 technologies have in common they need one or more of the following either to be extremely scaleable being able to address very high throughput terrible to have very low latency so if you are an autonomous vehicle you want the decision to be taken in a very timely manner it's part of your important. One which is very often overlooked is we need to deal with big data but what big means large and not something else but far too often I hear big data is not about large volumes of data I don't believe you can say that it's about big volumes of data and it is also about energy efficiency if you think about wearable devices whether it's a mobile device whether it's wearable devices whether it's a Google glasses whether it's wearable computer business say we need to be energy efficient if you look at all those sensors in an industry context say we need to be energy efficient why? because you want them to last as long as possible with a charge or maybe you want them to use environment or solar energy or whatever to last if they are in a hostile environment but what's interesting is that all these capabilities how do you achieve that is what we call programming for performance right? and ultimately these are the high performance computing techniques I've got here a few of my colleagues Michael Lyser who is hitting about novel technologies area and he's a man who does all the interesting work I'm just doing the talking which is far more interesting but we've got very exciting collaborations going on with the likes of Intel, Nvidia, Data Diode Networks Cylins and so on and I can give you a few examples it's a medical device we brought down a diagnostic system from 24 hours to only 4 minutes right? so now we've got back then a big medical device company in our funding actually the work with Enterprise Ireland to allow us to port this software for one of the medical device prototypes we did similar achievement with the UK financial services company again we reduced risk calculations from 4 minutes to 1 second so there are huge applications for your very broad range of domains the second part is to look at opportunities for Ireland because I think it's nice to be not to be sitting on the fence, it's better to really be part of the action in my opinion we're living from significant and unparalleled opportunities but we need to be very agile as I said we cannot afford the luxury of thinking about it for 2 years and we need a multidisciplinary approach and I will explain later what I mean by this new technology we're going for an era of digital transformation and in my opinion it's due primarily to the convergence of all these key technologies at the same time we look at the cloud computing the internet of things, the big data the analytics, the machine learning the HEC, you put all that together all this capability together to solve problems it's amazing amazing what you can do nowadays and where the discussion recently we coordinated a project on smart agriculture and on flood prevention flood management making a combination of earth observation from satellite data from high resolution computer modelling and from sensor data on the ground it is absolutely unbelievable what you can achieve nowadays in terms of precision agriculture and I think whoever makes the first move of that we will draw major, major areas but the other thing I wanted to emphasize is the growing role and the growing importance of technologists so I'm not saying researchers researchers are extremely important but researchers alone will not bring you the finite, the full package of the full solution you need to get technologists involved in exploiting this technology and last but not least we have an extremely favourable policy environment and particularly at the level of the European Commission so some of you would have noticed last week we had the celebration of the 60 years of the Treaty of Rome and we had as part of this a digital day and that suppressed release on the European Commission and you will see that the Commission will bring ministers together tomorrow to discuss high performance computing connected mobility and industry digitisation we've gone a long way now to get HPC recognised as the number one technology here in the digital transformation agenda that you are you wish to have to look at it more closely you find a lot of tweet posts and blogs from Commissioner Heatinger from Dr General Robert of Eula from the Vice President of the EU it's actually a map of a number of European countries which have engaged in this industry for policy agenda and there is going to be an extra nine countries joining them shortly so there will be a total of 22 or 23 EU countries working together to get this digital transformation across this industry for policy it's very exciting so after everything I said earlier was one of the domains where I think Ireland has a good share of goals on computing and it is primarily thanks to friends from IDA Ireland because IDA has been a super addition to all these companies here in Ireland both the IBMs and Google, Microsoft and Intel and Accenture Labs are innovation labs as well there are partners from OneCubit and these are the people who are making this story who are making it happen so there is technology guys here and I didn't plot underneath but all the financial services manufacturing, all these other domains which are going to benefit the most from quantum computing are also located here in Ireland so it might seem to be very silly not to avail of that ecosystem to do something about it and these are the four application domains where we consider quantum computing and of course Ireland is one of the most respected centres internationally it is a mix-up service but I have to take this opportunity to plug the fact that we have been recognised among the top 5 centres in the world by Intel and LGDL with Xilinx who were the only academic centre to help their certification in design and Data Direct Network with a master of Singapore and Saudi Arabia so that's a truly and genuinely global partnership of those companies what demonstrated the ability certainly to pick up new technologies very quickly one final thing I need to say is that I wouldn't be able to produce all that material on quantum computing without a little help from Mark Kyle and his colleagues of Accenture but I promised that in exchange I would point out that if you are interested in quantum computing I've got the e-mail address of the person what we hope that document and that you should certainly contact them that are very interested in hearing about anyone who is interested in exploring quantum computing and what you can do for everyone else what Mark mentioned is that they are currently porting with their partner room in excess of 150 real-life test cases industry test cases on quantum computer so it's not science fiction, it's happening as we speak Dipler means another domain which recently got a lot of publicity to admit until a few months ago I just didn't know what it was but everybody was talking about it you know bidding somebody a plane go for the first time then I get to do something maybe a little bit more useful with some plane go which is to look at detecting eye disease I think it was a whole free hospital in London in NHS and I think that data acquisition for mobile phone app and for that again to determine to identify this and then somebody's item again I learned through that process I didn't know but it's that if you suffer from diabetes you are 25 times more likely to develop blindness but it's that if you identify that the problem is coming at a very early stage in 98% of cases you can actually you can control the situation I think it's fascinating so this kind of technology can actually save people's sight if used accordingly for detections I think it's worthy of of managing that it's a genuinely useful innovation now the second one is to look for the second collaboration that also was carried out with a different hospital in East London was a huge, looked at huge data sets again so here we make a comparison between Formula One and Deep Learning we say big data is a fuel with that big data you will go to where the last time the low precision neural network is a driver and this is your Formula One car this is a new generation of devices that utilises what we call low precision low precision communication and you see in the video on the internet moving news is worth of mention that's my good friend David Moroney who is the great great Irish entrepreneur, great success we got bought by Intel a few months ago the great Irish success so basically what you do you travel with huge amounts of data and your neural networks we discussed over lunch whether it was a black box or not is going basically to determine basically give you a verdict as to whether or not particular eye condition is as occurred what's interesting there is a map on contradiction between inexact computing here you spoke earlier about precision medicine precision agriculture and now I'm saying in fact computing it's a bit of a contradiction isn't it so what's happening in fact is that yes it isn't exactly the case of one data set because remember you possess huge volumes of data so statistically the result you get is actually a lot more precise and it should be for a deterministic approach and it means that because you are happy with exact results then you can afford to use only a small part of your processor so in the past we went to 32 bit 64 bit processor increasingly precise calculations now we go in the opposite direction and we got processors who are going to do floating point operation with commas floating point numbers with only 8 and that's perfectly appropriate for even fact computing and that kind of boost tomorrow because when you go say from precision to single precision automatically your device goes twice faster so you guys have boost for free basically just by switching to a lower precision technology so it's very exciting I was mentioning earlier the conversions of all this technology this is a good example the second one I feel very passionate about is precision medicine and again it's very interesting for us to remember how far we went and how quickly we went so the first sequencing machines 70 years I was told with the first machine in it that cost several billion to actually get one human genome several years 2000 with the first human genome project and what's been happening here with breakfast in terms of innovations in the creation of this high food sequencing machine we have been able to lower the cost so this is the cost 100 million, 10 million, 1 million 100k and so on so we are going to worse the 1000 US dollar genome it's increasingly close before human being it means we will be able to have the whole news of genome data as part of your medical treatment but it comes with a difference because when you get that graph it does represent the volumes of data generated by this technology how much storage do you need how much data is actually generated by those technologies and here you have three lines the blue line is more slow the yellow orange line represents the projection from Illumina who was a leader in the domain of high food sequencing and the red one is actually a historical trend historical trends say you double the amount of data, the volume of data every 7 months so give you a sense again of how much of requirements are going on I didn't need to mention that because when we go to our funders we will make a case for certain equipment and there's always a sense that the check has been signed, you've got your equipment just go away, you probably won't need anything else for the next 5 years I said well, not quite because technology goes so fast that equipment is actually becoming observesent within only 2-3 years that's unfortunate but as you see in precision medicine is interesting so many people again do not understand why HPC, why data management with a huge volume of data to process you need to get a very high level of processing capability and does your model, you went somewhere you downloaded the data, you were tested now it's finished, the data is too large to move is what you do, you move the computation where the data is but also we did realize that the way we did medicine in the past say you had routinely we prescribe say aspirate people with some kind of heart problems and so on so the type of treatment was based on average efficacy of the drug for the population and so dosage you received very much depends on the advice of the chief physical officer so what country you live in now we do a far more personalized approach with precision medicine for example you got a number, typical cancer have different strands, different mutations so the first thing you would do is to identify the genome for a particular tumor to see what kind of tumors are you actually affecting with then you look at the genome for the patient as well and with all those different analysis you are going to identify your therapy which is going to be optimal for the person, for the patient and hopefully with far more positive outcome but you need integration of infrastructure such as HPC data and lab equipment institutions from the clinicians and so on technologies and so on you also need data protection on the side which is very important and you can do these parties as well so I think that's a slide particularly particularly passionate so next one I go back to what I was saying earlier so I won't read it again because pretty much repeating what I was saying about these two collaborations between Google the first one is the Royal Free Hospital the second one is Moorfield High Hospital the only thing you would mention is that they have the largest repository of optical coherence, OCT, tomographic data scans in the world and they share it with Google I am told about a million animals digital eye scan with related information about the course of treatment and whatever was relevant information about the patient information but that supports in terms of the ability of legislation to keep up and the first one in part of Clarco's lot of controversy and people who were openly criticizing the press the right of the NHS the right of the hospital to share this data with Google I think there is a natural anxiety as a prospect of private data being in the hands of big black multinational corporations most of us are a bit worse but fearful about that why I say do it first and this was a response at the time that was given by the hospital they just followed standard policy in terms of sharing data with the party organizations now is that appropriate or not what's very interesting is that for the second collaboration the data were anonymized and there was no such public protests in terms of Google work in terms of hospital work but what's very interesting is how to strike the balance between the right to privacy and the public interest and in the UK there is a very interesting report because the UK can pick up principles and I like very much the 7th, 4th and 7th principles which is the duty to share information from the other people as a duty to protect patients from dementia now I don't know too much about this so if you have any questions about Cardicot that's my colleague Emma here in the corner is looking after our public sector collaborations works very closely with the central statistics of this and with the health research board Emma could tell you all about this principle what I wanted to point out and passing however is that the function of what we call this preceptor party data processor function in Scotland is actually fulfilled by their national high-performance marketing centre where I used to work so again it demonstrates the validity of these approaches and these are my final two slides because I realise it's getting warm and everybody is feeling a bit like an English sister we need to be strategic I said please so that's not a requirement just hope are we ready? yes we do have this platform absolutely we have innovation driven policies it goes without saying I think innovation 2020 demonstrates the importance of innovation in national policy and we have very strong industry links both for the remarkable work of idea Ireland and enterprise but in my opinion there are a number of key structural issues and the way I see it the venture with prioritisation based policies it can lead to a silo mentality or unnecessary silo of research and development that's part of the problem and also underfunding of platform science and technology and for me that's a key problem so then if you look at this emerging domain earlier quantum computing precision medicine and so on you need technologies to work hand in hand with domain specialists sometimes domain specialists for multiple domains so if you start putting in place a system that tries to partition, isolate and create silos in my opinion this is counterproductive what's even more counterproductive is to ignore the platform science technology and infrastructure and my second difficulty is the chronic under-investment in national infrastructure and I think what will be very useful here is to have some level of routine benchmarking of hybrid computing and data management infrastructure per capita compared to other European countries and I think that's a well known fact that we currently have 6 UK universities with more computational power for their own university individually than we have for the country as a world and I don't think that's a normal situation that I don't think this is causing a lot of difficulties to people who rely on processing power to carry out their business for example I think another example was the lack of availability for a platform for data management I think again the mystery to name is that this is certainly slowing down as you know from the observation of learning I can name some but a few most of the companies that come under name come to see me, the first thing they say is do you have very large data sets? why? because you want to do machine learning you want to do deep learning you don't do deep learning I was going to say I'm a floppy disk you did much larger data sets hundreds of terabytes of data pipes why do you put them if you don't have an infrastructure to store this data? I'd love to ask that question particularly with big data analytics in regard of the priority infrastructure is not just about equipment infrastructure is about software and more importantly is about people so in my view the funding makers should take care of addressing infrastructure as a whole including the funding cost including the software including the people all too often are on the side and I think the whole importance of technology should not be understood and I think if you want those guys like my colleagues here and effectively with domain specialist clinicians, whatever it is in industry or academia we need access to opportunities to funding but we also need clarity of resting and the problem with the academic world I hope I don't offend too many people but it is one of the last remaining caste systems it is a caste system right? so you are a career academic or you are the riffraff you are the technician a glorified technician but that's a problem because we don't form high performing teams without respect and that's the last slide key recommendations are taken to free I think we should consider focusing on a top-down approach to national infrastructure funding and I've seen somebody here from the department of the t-shop so maybe that's my one we need an approach which caters for cost department or cost agency initiatives because I think to my mind as taxpayer it would make sense to fund data management infrastructure computing power, a different one for agriculture different one for jobs, different one for education different one for each government department it just doesn't make any sense this needs a cost department role and you achieve considerable economies of scale not only in terms of the hardware but the software infrastructure in terms of deployment and operation so we need that big deal and the management department recently visited the data center of the revenue commissioners and I was delighted to see that all the equipment helps the little batch from the department and you find pretty much in the single government department you find the it was the prism services you had revenue commissioners all these state agencies together understood the benefit of the data center and sharing the same infrastructure so there is a good understanding of that domain of the benefits of connecting the doors and having a cost department approach so in such infrastructure to be robust near machine critical and loyalty disciplinary robust why? because if you want to use it to deliver public services if you want to use it as a platform collaboration of industry it need to be 24x7 if you look after it it's just not good enough so we've been operating the national weather forecasting for over 10 years we know how to do this kind of business we are very happy to help people who are offered this opportunity disciplinary at two different levels because the current approach tend to lead to organization of resources again as I said if you start distributing infrastructure funding across several universities because you are concerned with your popularity of the university sector those are genuine benefits of bringing together data cells from various disciplines within the same infrastructure greatly facilitates you look at transport weather or climate and health and so on you can look at clear correlation between say climate and the development of certain conditions like respiratory disease and so on or transport for example if you start by putting them all over the place flooding is a good example right? flooding seems a calculation once you need almost I think about 10 different data sources for an efficient flood modelling system I mean again if you all decide under the same infrastructure it's much easier to deploy models the second one if you need to introduce programs to support HPC technologies again it's a bit self-serving I apologize for that but I think it's a good platform to I hope you allow me a bit of self-serving promotion but there is a significant opportunity for the country in terms of training education there is going to be huge demand before too long for people who know how to deploy these kind of technologies very effectively and again that's something not only to look at the third level sector but also something to look at the younger age and I think there are some very good initiatives in this domain I heard about some initiatives from Google earlier over lunch so thank you for that so it's a kind of initiatives we are looking to see adopted and the need to bring despite your esteem and come back to it again which is maybe reconsidering certain eligibility criteria that places us outside of training models and the way in which applications are being reviewed and the final one I think we have a need to expand processes we can always do better no matter how good what we do is we always need to do better and my feeling in terms of policy implementation is that we could greatly benefit from the creation of an international strategic advisory network participating in such groups in the UK for a good few years and that allows me to show that whatever awareness of technology is going to be found and we need to be responsive what's happening in other countries that are connected to the rest of the world and not taking our decision in vacuum and the final one really is I would also be very keen on the position of chief technology officer to the government being created we saw the benefits of creating the position of CIO as no doubt we all understand the benefits of what a good chief scientific advisor can do in that domain as well but I think technology is sufficiently if some domain is sufficiently distinct from science in my view that it deserves to get somebody really in charge of this brief and somebody I would add with a long track record in industry and a long track record of innovation would be even better if we could secure as a participation in such a person Hop, I wasn't controversial Well, yes you were Thank you very much