New Delhi: Computerised draw of lots for selecting pilgrims for the Kailash Mansarovar Yatra was held here Wednesday, in which for the first time, preference for fresh applicants or route choice of senior citizens was incorporated into the software, MEA officials said.Foreign Secretary Vijay Gokhale presided over the draw of lots held at the Jawaharlal Nehru Bhawan, and also urged pilgrims to help “protect and preserve” the fragile environment in the Himalayas. Also Read – 2019 most peaceful festive season for J&K: Jitendra SinghThe ministry’s consistent effort has been to get as many applications, because it thinks this Yatra is important not only for those going for spiritual or religious purposes, but also to build relations between India and China, he said. Ministry of External Affairs organises the Yatra during June to September each year through two different routes – Lipulekh Pass (Uttarakhand),and Nathu La Pass (Sikkim). Known for its religious value and cultural significance, it is undertaken by hundreds of people every year. Also Read – Personal life needs to be respected: Cong on reports of Rahul’s visit abroadFor Kailash Mansarovar Yatra 2019, the ministry received 2,996 applications, out of which 2,256 are male applicants, and 740 females. As many as 624 senior citizens had applied for the yatra. For Lipulekh route, there are 18 batches with 60 pilgrims per batch and for Nathu La, 10 batches with 50 yatris per batch. Two liaison officers will assist each batch of yatris. “It is our hope and endeavour that we provide the first chance to those who didnt ad the chance to go to the Yatra, the first-time applicants and we give priority to them. And, of course, to senior citizens as well,” Gokhale said. He said, consistent efforts have been made to make the Yatra website as pilgrim-friendly as possible. “We also have a helpline now for applicant yatris. And, emails received from them are regularly monitored and applicants are suitably advised in a time-bound way,” he added. A senior official said, the Nathu La route is less arduous compared to Lipulekh route and so preferred by senior citizens. “Earlier they would tell us their route choice while applying for the Yatra, and we would try to accommodate their request in the overall process. This year, we have incorporate the preferences of first-time yatris and the senior citizens in the algorithm of the software used for the draw of lots, which is for the first time,” he said. The selection is a fair computer-generated, random gender-balanced selection process, the MEA said in a statement, adding, the selected yatris are informed through mobile text messages and email. “Since 2015, the entire process commencing with on-line application till selection of yatris is a fully computerized process. Therefore, applicants do not need to send a letter or fax to seek information. The feedback options on the website can be used for obtaining information, registering observations or suggestions for improvement,” it said. The Yatra involves trekking at high altitudes of up to 19,500 feet, under inhospitable conditions, including extreme weather, and rugged terrain, and may prove hazardous for those who are not physically and medically fit. Gokhale urged yatris to strictly observe safety norms, for themselves and also for their fellow pilgrims. The Yatra is organized with the support of the state governments of Uttarakhand, Delhi, and Sikkim; and the cooperation of Indo Tibetan Border Police (ITBP). The Kumaon Mandal Vikas Nigam (KMVN), and Sikkim Tourism Development Corporation (STDC) and their associated organisations provide logistical support and facilities for each batch of yatris in India. Sushama Katarai was delighted after the draw, and said, “I have always wanted to do this Yatra. It is a like a dream come true for me.” Varun Khandelwal, a Delhi-based pilot, who was put on the waiting list after the draw, said, “I am hoping I will make it eventually. I have always wanted to experience Kailash Mansarovar Yatra.”
LONDON — The Latest on Britain’s exit from the European Union (all times local):5:45 p.m.British Prime Minister Theresa May has suffered an embarrassing parliamentary defeat on Brexit as lawmakers remain resistant to her EU divorce plan.The House of Commons voted 303 to 258 on Thursday against a motion reiterating support for May’s approach to Brexit — support expressed by lawmakers in votes just two weeks ago.The defeat is symbolic rather than binding stresses how weak her hand is as she tries to secure changes to her divorce deal from the EU in order to win backing for it in Parliament.The government was defeated when the members of a pro-Brexit faction in the governing Conservatives, the European Research Group, abstained because they feel the government is effectively ruling out the threat of leaving the EU without a deal.May is struggling with little sign of success to win backing for her deal from both pro-Brexit and pro-EU lawmakers in Parliament, which rejected the agreement by a whopping 230 votes last month.The U.K. is due to leave the EU on March 29.___9:25 a.m.Britain’s prime minister is scrambling to avoid another defeat on her Brexit strategy amid opposition from members of her own party who fear she is moving in the wrong direction in efforts to overcome the impasse blocking a deal.Hard-line pro-Brexit lawmakers say a measure to be voted on Thursday rules out the threat of leaving the European Union without an agreement on future relations, undermining Britain’s bargaining position. Prime Minister Theresa May has previously ruled out a “no-deal” Brexit as she attempts to win concessions from the EU after Parliament rejected her deal last month.European Council President Donald Tusk has reiterated his frustration with Britain, tweeting “No news is not always good news. EU27 still waiting for concrete, realistic proposals from London on how to break #Brexit impasse.”___Follow AP’s full coverage of Brexit at: https://www.apnews.com/BrexitThe Associated Press
Studio manager Fiona Furness was one of the first to trial the game and has lost two stone in weight so far, dropping from 11 stone to nine stone. Mrs Furness, who is in her 50s, said the “pounds just melted way.”“The results have been remarkable,” she said. “I used to feel really guilty about my bad snacking habits. I’d often be rushing about, and I’d grab something high calorie and unsatisfying – often a packet of crisps. I’d be hungry again really soon afterwards so it became a vicious cycle. “These days, if I am feeling peckish I’ll go for a banana or a pack of almonds. That’s the food I’m craving. I’m now closer to nine stone than 11 – the pounds just melted away over eight or nine months without me even noticing. Want the best of The Telegraph direct to your email and WhatsApp? Sign up to our free twice-daily Front Page newsletter and new audio briefings. The game works by flashing up pictures of healthy and unhealthy food and the user has to react by only pressing on the healthy foods to score points. The simple act of ignoring unhealthy foods, and stimulating the reward response to healthy foods is enough to retrain the brain into craving healthier options, say scientists. A study of 83 adults showed that people who played the game online just four times in one week lost weight and ate an average of 220 kcal less per day – roughly equivalent to a chocolate-iced doughnut. “It’s very exciting to see that our free and simple training can change eating habits and have a positive impact on some people’s lives,” said Dr Lawrence.“In an age where unhealthy food is so abundant and easily available and obesity is a growing health crisis, we need to design innovative ways to support people to live more healthily.”We are optimistic that the way this app is devised will actually encourage people to opt for healthy food such as fruit and vegetables rather than junk food.” Prof Natalia Lawrence “The weight loss wasn’t really my goal though – I feel younger and more energetic. Perhaps I’m particularly susceptible to this kind of brain training, but it has been transformative for me.” Nearly two thirds of adults in Britain are overweight or obese, and the UK is forecast to be the fattest country in Europe within 10 years Users of the app, who should ideally use it for a few minutes a day without distractions, can tailor it to reduce compulsions to unhealthy food they have most problem with, as well as alcohol, but not to reduce consumption of healthy foods including vegetables. The scientists have launched a crowd-funding campaign to raise up to £5,000 to develop the app, this week made available for Android devices, into an app that can be used on iPhones and iPads. A game which trains the brain to avoid unhealthy food such as cakes, biscuits and chocolate can lead to ‘pain free’ weight loss and cut energy intake by more than 200 calories a day, scientists have said. Psychologists at the University of Exeter showed that less than ten minutes a day of ‘brain training’ using a game which takes away the ‘mental reward’ of sugary and fatty foods, reduces calorie intake. Professor Natalia Lawrence’s Food Trainer app is free and is being launched this week on Android devices.It is based on neuroscience research which suggests people are more inclined to choose fatty and sugary foods because they activate the brain’s reward system, stimulating the release of dopamine and endorphins, which can produce feelings of pleasure and make the person want more.
Share this:Click to share on Twitter (Opens in new window)Click to share on Facebook (Opens in new window)RelatedGECOM FIASCO: Opposition, Govt Attorneys to meet next weekFebruary 2, 2017In “Politics”GECOM Chair: New list will be submitted soon- TeixeiraMarch 29, 2017In “latest news”PM contradicts President’s position on nominations for GECOM chairmanshipJanuary 30, 2017In “Politics” … AG unprepared, requests more time for interpretation of Constitution – NandlallThe much anticipated meeting between Government and Opposition representatives to iron out the interpretation of the Constitution with regards to the appointment of a Chairman of the Guyana Elections Commission (GECOM) got underway on Wednesday and was proven to be unfruitful.Attorney-at-Law Anil NandlallThe Opposition appointed Attorneys – Priya Manickchand and former Attorney General Anil Nandlall – met with Attorney General Basil Williams SC, following a proposal made by President David Granger to Opposition Leader Dr Bharrat Jagdeo, for legal minds from the two sides to meet and discuss the appointment of a GECOM Chairman.When contacted on Wednesday, Attorney General Williams told Guyana Times that during the meeting, he heard submissions by Nandlall and asked for some time to review the submissions in order to arrive at a proper conclusion.“I spoke with Mr Nandlall respectfully, he made his contention and he cited authorities; and I said I’ll take what you say safely by reading the authorities and then we’ll respond to his submissions,” Williams stated.Attorney General Basil WilliamsIn a statement released later to the media, Williams posited “I could have indicated to him when you look at constrained, the provisions, one: “there shall” – it’s mandatory it tells that you must first look at the mandatory provisions, which deal with descending order: judge, retired judge, person who could be appointed a judge, who is qualified to be a judge and the …any other fit and proper person and so I could have done that and said, ‘Well I don’t agree with your interpretation but I’m doing better than that because it’s not a question of Mr. Nandlall and I. It’s a question of the people of Guyana and the question of the proper law to be applied for the future in relation these matters.” Meanwhile, in a statement to the media, Nandlall disclosed that during the meeting, the Opposition lawyers proffered their interpretation of Article 161 of the Constitution, in writing, and supported their position with a number of case law authorities from Guyana, the Caribbean and the Commonwealth.However, Nandall pointed out that the Attorney General was “unprepared” to put forward his or the Government’s interpretation of Article 161 of the Constitution, despite several requests for him to do so.“Instead, he indicated that he will need time to interpret our contentions and prepare his response. All of the interpretations offered by us (Wednesday) were fully and publicly ventilated in the media… I am disappointed by the lack of preparedness of the Attorney General, which resulted in nothing tangible emerging from the engagement,” the former AG asserted.Nandlall went on to say in his statement that “Quite frankly, I was hoping that the Attorney General would have been ready with his position on the matter today (Wednesday); that may have resulted in this matter being concluded with dispatch and decisively. In the meanwhile, Guyana’s democracy continues to hang in the balance.”Moreover, Nandlall stated that the Attorney General could not identify a date for a second engagement on the issue, which the Opposition considers a matter of great national importance. “The meeting ended with the Attorney General being unable to identify another date available in his diary for us to meet again,” he noted.Last week, President David Granger said he is anxiously awaiting the outcome of dialogue between the two sides.“I’m anxious to move ahead, it’s a constitutional office… I have asked them to meet and see in what ways we can have a common interpretation of the Constitution. In the final analysis, it’s a constitutional matter and I felt that the Constitution was quite clear so if there was any opaqueness or obscurity, I hope that the meeting between the two Attorneys would clarify that,” the President asserted.The Head of State further maintained that he was correct in his interpretation of the Constitution that the nominees for GECOM chairmanship must be a judge or have qualifications equivalent to that of a judge.“…as I said, my interpretation of the Constitution was quite clear. I’m not saying that the person has to be a judge, but if you read the whole article of that Constitution, it intends that the person must possess certain qualities and we are looking for those qualities,” he posited.The controversy on the matter revolves around the different interpretations of the Opposition Leader and the Government on Article 161 (2) on the appointment of a new GECOM Chairman.The Opposition’s understanding conforms to the updated Constitution which incorporates the Carter formula to have a democratic process for the appointment of a chairman.Granger’s interpretation, on the other hand, seems to reflect the old 1980 Constitution which limits the pool of persons to be appointed to GECOM’s helm to only judges or those eligible to be a judge.
The President of the Austrian Parliament, Barbara Prammer, has honoured Alan R. Hill, President and CEO of Gabriel Resources, the ‘Responsible Manager of the Year’ Award in the international category at an event hosted at the Austrian Parliament in Vienna.The prize is given to those who excel in responsible management as part of the ICON-Vienna Congress, which showcases trends and innovative developments in Europe. ICON-Vienna established the CSR prizes in the categories of international, public interest, small and medium sized companies, middle management and top management. Members of the ICON-Vienna Conference include Trade Union Austria, the Association of Industrialists, the Chamber of Commerce and the Ministry of Economics.Hill was recognized for his leadership at the helm of Rosia Montana Gold Corp, a large-scale project that could become Europe’s biggest gold mine. According to the jury, the award was given to him because he is committed to social and ecological responsibility in a challenging business area like mining, which does not typically apply such criteria. The jury further stated that this example of setting up a mine in Romania could serve to pave the way into the future for the entire industry.As an industry veteran, Hill has spent over four decades in the mining industry, leading projects all over the world, and has seen first hand how countries’ natural resources can translate into wealth for nations and people. For the last two years, since he became President and CEO of Gabriel Resources, Hill has worked toward meeting the challenge of his latest project: to build a mine to the highest industry standards, as well as use the project as a catalyst for regional sustainable development.Under Hill’s leadership, Gabriel Rosia Montana has pioneered social programmes and civil society partnerships in the areas of education, health, environment and business, as well as introducing company corporate governance policies built on openness and transparency. Hill has ensured that his company applies the highest standards to all of its operational areas.Hill’s initiatives include:Introducing open and transparent management practicesEnlisting top specialists to help design a project fully compliant with EU and international standardsInvolving leading Romanian and international experts in the assessment of the relevant environmental impacts of the proposed projectEncouraging active environmental protection measures, such as forestry programmes, environmental awareness and education actions and partnershipsEstablishing an ongoing social programme which helps disadvantaged people in the project areaHelping set up a professional development training programme for the local community, through which over 600 people have been trained in areas directly related to the core business of the companyDesigning a future village to combine modern facilities with traditional architecture for the local community to offer them significantly higher living standardsLaunching of the Rosia Montana MicroCredit, a micro-lending institution that was set up recently in order to encourage diversity and development of the local economy through customised financial solutions.Gabriel Rosia Montana currently employs about 400 people, most of them from the local community, in an area experiencing more than 70% unemployment. Once operations have begun, it will employ over 1,200 people for two years of construction, 600 during the 17 years of the life of the project and will likely generate over 6,000 indirect jobs. The company has signed protocols with local authorities to ensure preference of employment is given to locals from the project area and surrounding communities.
A LETTER WRITTEN by the former president of the ECB to the then-Irish Finance Minister Brian Lenihan could be published – weeks after a complaint from the European Ombudsman.The 2010 letter was sent from Jean-Claude Trichet to Lenihan but its release was blocked by the Governing Council of the European Central Bank (ECB).On 7 March, European Ombudsman Emily O’Reilly said she was unhappy that the ECB had blocked the release of the letter.She said she regrets that the Governing Council “has wasted an opportunity to apply the principle that, in a democracy, transparency should be the rule and secrecy the exception”.RequestToday, Sinn Féin said that its MEP Martina Anderson had submitted a formal question to the ECB asking under what conditions they will release the letter and what action they may take if Minister Noonan released it unilaterally.She put a motion to the European Parliament which called on the ECB “to publish the letter of 19 November 2010 from Jean Claude Trichet to the then Irish Finance Minister as requested by the European Ombudsman”.The motion was accepted.Read about the letters written by Lenihan that have already been releasedSinn Féin’s Midlands Northwest EU candidate Matt Carthy criticised other MEPs for their approach to the vote. He said that Jim Higgins of Fine Gael voted against Anderson’s motion, while Pat “The Cope” Gallagher of Fianna Fáil “didn’t vote”.LettersGavin Sheridan of TheStory.ie said that he had submitted a request to the ECB for all letters sent to Brian Lenihan or his office in November 2010. He said he was told that the release of one specific letter’s contents would “undermine the protection of the public interest”.At the time, Sheridan said he intended appealing the decision.Read: Ombudsman unhappy ECB won’t give Brian Lenihan letter to journalist>
Facebook Twitter: @NeosKosmos Instagram The Victorian Electoral Commission officially declared Jennifer Kanis the winner of last Saturday’s by-election on Thursday morning.“It’s terrific,” Ms Kanis tells Neos Kosmos on being elected, and says she is “really honoured and pleased to be elected”.Ms Kanis edged out Cathy Oke – the Greens candidate – by 1067 votes on a two-party preferred basis.After a gruelling campaign, Ms Kanis says the hard work is just about to start for her.“The first thing I am looking to do is to continue what I’ve been doing over the last couple of months which is talking to people in the electorate and and meeting with the various community groups,” she says adding although she knows her electorate well, she wants to get to know them better. And by doing so, she will be able to represent their views in parliament and make sure the policies that they put in place align to Labor values and with what people want in Victoria. Ms Kanis said as far as her projects go she is going to concentrate on holding “Ted Baillieu to account”. “There are acouple of things that he’s done that people are unhappy with – the TAFE cuts for instance … we are going to make sure that Ted Baillieu and his government know just how unhappy people are about cuts to education.”The win comes as somewhat a relief for Prime Minister Julia Gillard, as a loss would have added to the pressure of a leadership rumblings in the party. Added to that, the pressure that Labor has held the state seat of Melbourne for more than half a century.
Belgium boss Roberto Martinez has made Eden Hazard’s future at Chelsea even more uncertain by suggesting that now may be the right time for him to leaveHazard has been strongly linked with moves to both Real Madrid and Barcelona recently with the winger himself having hinted that the prospect of ending his six-year spell at Chelsea for a move to Spain may be inviting.Having rejected a new contract and with just two years left on his current one, Hazard may be able to secure a move away from Stamford Bridge this summer and it appears that he will have Martinez’s blessing.“It could be the best time to try something different,” Martinez told radio show El Larguero, via ESPN.Chelsea hat-trick hero Tammy Abraham hopes for more Andrew Smyth – September 14, 2019 Tammy Abraham hopes this season will be his big breakthrough at Chelsea after firing his first hat-trick for the club in Saturday’s 5-2 win at Wolves.“He is a player who has the maturity and a lot of leadership. His play is based on talent. Hazard could carry a new project anywhere in the world. He is at the best moment of his career.“For me, it’s very easy to answer that question. He could fit in any team in the world. Players need new challenges and new projects. Perhaps it’s a good time for Chelsea and for Eden [to split]. I would be very surprised if Chelsea didn’t have big offers for him right now.”Hazard ended his superb World Cup campaign in style on Saturday by netting the second goal as Belgium secured a third-place finish after beating England 2-0.The 27-year-old scored a total of three goals and added two assists in his six games at Russia.
Genoa have confirmed they are yet to receive offers from reported Real Madrid and AC Milan target Krzysztof Piatek.The Polish striker has found the net a remarkable 19 times in 21 appearances for the Italian club this season, having joined from Cracovia Krakow last summer.The Serie A side paid only €4.5m for Piatek – who has also been linked to Barcelona – in that deal but his market value has multiplied with reports suggesting he would only be sold permanently in a move in excess of €60m.A report cited on Football Espana speaks of ‘concrete interest’ from the European giants while Atletico Madrid and AC Milan are equally monitoring the situation.Zidane reveals Sergio Ramos injury concern for Real Madrid Andrew Smyth – September 14, 2019 Zinedine Zidane has put Sergio Ramos’ availability for Real Madrid’s trip to Sevilla next weekend in doubt after withdrawing him against Levante.“We are waiting for offers for Piatek, there is nothing official yet,” Genoa CEO Giorgio Perinetti told RMC Sport.“Milan know that we would like to keep the player until June, but we will evaluate any proposal that may come in.“It is not our intention to sell Piatek now and we cannot wait until the last day of the market.”
Something for the weekend: Everyone knows that parents can bring an abundance of new skills into the workplace when they return from leave, but what if their childcare and home-keeping responsibilities were able to earn them a wage?According to a new tool from greeting card organisation Funky Pigeon, released in time for Mother’s Day on 31 March 2019, London-based parents could be worth a whopping £263,587 a year for completing their day-to-day tasks around the home, while mums and dads living in Bristol perform roles that could earn £185,607 a year.The tool is based on eight tasks that parents perform while raising children; matched with the closest professional equivalent.For example, time spent cooking can be compared to those working as a chef, hours cleaning could equate to a cleaner’s pay, while driving chores relate to the role of a taxi driver. Helping children learn puts parents on a par with teachers, organising the family and household is similar to a professional personal assistant and tackling mountains of washing fits with the responsibilities of working in a launderette. Tending to their child’s physical health, for example, could be comparable to being a nurse, and looking after the family’s mental wellbeing is linked to a psychologist’s job role.Funky Pigeon then collated the wage and salary data for these comparable job roles across 1,000 UK towns and cities.To use the tool, parents can input the number of hours a week they spend completing the eight different task categories. This data is then converted into a yearly salary for each job role, which is then added together to create a total parent salary.For example, if a parent spends seven hours a week cooking for their family, over the year that would equate to 364 hours of cooking. If the average wage for a chef living in the same area is £15 an hour, then the estimated salary for this one job would be £5,460.Here at Employee Benefits, we took a stab at using the tool for ourselves and managed to achieve a parent salary of £70,505 a year, as we are London-based. For working parents, we are sure this extra income would be very much appreciated after all their hard work…
Addis Ababa : The chief of staff of the Ethiopian army was shot dead during a coup attempt in the northern Amhara region, the Prime Minister’s office announced on Sunday on state television. General Seare Mekonnen died after being shot while trying to prevent the coup attempt on Saturday night, in which another military chief, General Gezai Abera, also died, the office said on state television ETV. Also Read – Shahid Afridi joins ‘Kashmir Hour’ in military uniform Advertise With Us The president of Amhara region, Ambachew Mekonnen, as well as his adviser, Ezez Wasie, were also shot dead in their office in the regional capital, reports Efe news. The coup against the government of this region, the second largest ethnic group in the country, began late Saturday in the capital, Bahir Dar, and was thwarted soon after by the federal security forces, the prime minister’s spokesman, Nigussu Tilahun, announced on ETV. Also Read – EAM Jaishankar calls on European Parliament President David Sassoli Advertise With Us “The coup attempt in Amhara regional state is against the constitution and is intended to scupper the hard-won peace of the region,” the prime minister’s office said in an initial statement. “This illegal attempt should be condemned by all Ethiopians and the federal government has full capacity to overpower this armed group,” it added, without specifying who was behind the attempt.
Popular on Variety Earlier in FilMart, the company also screened “Toto,” a Filipino comedy from director John Paul Su, on which it is handling global sales. The film follows a young hotel worker who schemes to attain a U.S. visa at all costs. It will be released on iTunes in the second quarter of the year. ×Actors Reveal Their Favorite Disney PrincessesSeveral actors, like Daisy Ridley, Awkwafina, Jeff Goldblum and Gina Rodriguez, reveal their favorite Disney princesses. Rapunzel, Mulan, Ariel,Tiana, Sleeping Beauty and Jasmine all got some love from the Disney stars.More VideosVolume 0%Press shift question mark to access a list of keyboard shortcutsKeyboard Shortcutsplay/pauseincrease volumedecrease volumeseek forwardsseek backwardstoggle captionstoggle fullscreenmute/unmuteseek to %SPACE↑↓→←cfm0-9Next UpJennifer Lopez Shares How She Became a Mogul04:350.5x1x1.25×1.5x2xLive00:0002:1502:15 Online rights sales platform, BidSlate will Wednesday give an offline world premiere to Emmy-nominated documentary-maker Sally Rowe’s “Old Dog” at Hong Kong FilMart.“Old Dog” tracks New Zealand farmer Paul Sorenson who has used 40 years of experience working with sheep dogs, to develop smarter and more intuitive training methods for fellow farmers. He also grapples with memories of a difficult childhood.Producers include Rowe (“A Matter of Taste: Serving Up Paul Liebrandt”,) BidSlate president and co-founder Roland Rojas, Regina Sobel, Benjamin Breen and Alan Oxman. ““Sally brings a discerning eye and unique POV to this quintessentially Kiwi story that’s certain to resonate with audiences the world over,” said Rojas.BidSlate is a platform offering buyers and rights owners a 24/7, secure and transparent marketplace to acquire or sell exclusive new content across any platform. It recently expanded its content offering through deals with Filmhub and Digital Media Rights.
Like Doctor Who, vinyl records made a major comeback in the noughties.Now, after the success of last year’s Record Store Day exclusive, Genesis of The Daleks, Demon Records returns for a limited 2017 run of Doctor Who and The Pescatons.The Sunday exclusive, available on 180g heavyweight vinyl, was originally released on LP and cassette (remember those?) in 1976.Written by Victor Pemberton, author of Second Doctor adventure Fury from the Deep (the episode that introduced the sonic screwdriver), The Pescatons features the voices of Fourth Doctor Tom Baker and Sarah Jane Smith, played by the late Elisabeth Sladen.The two-part play is the first audio drama based on Doctor Who, and remained the last until the late 1990s, when Big Finish revived the concept. To celebrate the return of @bbcdoctorwho today’s countdown to #RSD17 is Doctor Who & The Pescatons. @doctorwhosite #DoctorWho @WhovianNet pic.twitter.com/gtVQPAhgVS— Demon Music Group (@DemonMusicGroup) April 15, 2017The Doctor and Sarah Jane investigate a meteorite the Time Lord believes to be a Pescaton ship, belonging to the race of deep-water sharks. When one emerges from the Thames and makes its way to the London Zoo in search of salt water, the Doctor confronts it and the Pescaton dies. But that night, more “meteorites” land in the river.Surprise: London is invaded. The Doctor distracts the Pescaton with a “Hello, Dolly!” song-and-dance (not the best plot twist for an audio play), and the creatures retreat. The Time Lord, meanwhile, builds a high-frequency sound trap inside a sewer, lures Pescaton leader Zore to be destroyed, and the planet Pesca disintegrates, ending the city-wide assault.Doctor Who and The Pescatons was re-released on CD in 1991, and again in 2005; the latter, available via Amazon for $84.99, comes with an exclusive 45-minute bonus interview. A novelisation by Pemberton was published in 1991 by Target Books.Classic fans and hipsters, however, can visit a local participating record store on Sunday for a chance to snag the two-LP reissue on Pescaton-green colored vinyl. As an added perk, the gatefold release comes backed with sound effects on orange vinyl.“With the original artwork for each release back-to-back, this is a must for any Doctor Who fan,” Demon Records said in an announcement.The albums are limited to just 3,000 copies worldwide. Check out the full track listing below.Side A: Doctor Who and The Pescatons Part 1Side B: Doctor Who and The Pescatons Part 2Side C: Sound effects1. The Central Control Room In Exillon City2. The Dalek Control Room3. Metebelis III Atmosphere4. Styre’s Scouting Machine5. Dalek Hatching Tanks On Skaros6. Zygon Spaceship Control Center7. Sutekh Time Tunnel8. The Interior of XoanonSide D: Sound effects1. The Shrine of the Sisterhood of Karn2. Kraal Disorientation Chamber3. The ManDoctoragora Helix4. Atomic Reactor Runs Wild5. Wind-Mine Machine6. Distillation Chamber7. Cloning and Miniaturization Process8. Inside Doctor Who’s Mind9. TARDIS Interior (in flight)10. TARDIS Interior (stationary)11. Observation Screen Operates12. TARDIS Door Opens13. Sonic ScrewDoctoriver14. Fission Gun15. Tech Gun16. Gallifreyan Staser Gun17. Vardan Gun
Overclocking has been a way of getting more performance out of PC components for decades. It used to be purely applied to processors. You add more cooling and clock the chip frequency higher in order to gain more performance. Then the same thing happened with RAM and graphics cards. Overclocked versions of these components are even sold new with a price premium. Now overclocking is being offered on a monitor.If you think about it, there’s not much available to overclock on a display. You certainly can’t increase the screen size or go past the maximum resolution offered. Maybe you could speed up the interface menu changes? However, Acer has found one legitimate way to overclock its new Predator XB1 monitor, and that’s by tweaking the refresh rate.The Predator range of monitors is squarely aimed at gamers. In the case of the Predator XB1, it uses a 24-inch TN panel with a resolution of 2560 x 1440. It supports refresh rates up to 144Hz as standard, but Acer has included an overclock feature allowing you to push it up to 165Hz if desired. And if that video above is anything to go by, Acer is aiming to hit 180Hz eventually.What’s the benefit of this overclock? Well, that depends on how good the pixel response times are and the quality of the digital processing circuitry Acer has opted to include in the display. You’d have to assume it’s good enough otherwise such a high refresh rate would introduce image distortion or artifacts while you play. The specs suggest it is, with response times listed at 1ms.Anyone considering the XB1 as their next gaming monitor will also be pleased to hear Acer has included support for Nvidia G-Sync, meaning you’ll have a screen tear-free experience assuming you also have a G-Sync enabled graphics card in your rig. You also get a USB 3.0 hub, 2W speakers, 100% sRGB support, 350Nit brightness, and power use of just 27W. The cost? $499 with immediate availability.
Darjeeling: A leopard that had been tormenting the villagers at Lower Kaijaley in the Darjeeling Hills was finally trapped on Thursday.Villagers of Kaijaley, 2 km from Bijanbari, are spending sleepless nights with leopards taking away their dogs, goats and pigs. “It has been more than one and a half months that two adult leopards have been sighted in the Lower Kaijaley area. There is a small forest near the Reling Khola (river) which the leopards inhabit. They come to the village in search of easy prey,” stated KB Wattar, a resident. Also Read – Rs 13,000 crore investment to provide 2 lakh jobs: MamataAccording to Wattar, the leopards have killed most of the dogs of the village and nearby areas numbering hundred. Then they switched to goats. The two have killed more than 40 goats and pigs also. Finally the villagers had given a deputation to the forest department. The Forest department had set up a leopard trap in which the leopard was trapped. It had killed the goat that had been used as bait. “We had received information from the villagers and accordingly the range office had set up a leopard cage on Wednesday. On Thursday morning a female adult leopard was found trapped in the cage. We will take the trapped leopard to Bengal Safari in Siliguri,” stated Vikas V, Divisional Forest Officer (DFO.) The villagers have requested for more traps to trap the other leopard also. Bijanbari is located at a distance of 30 km from Darjeeling town.
The video will start in 1Cancel Play now StokeonTrentLive – we’re #localandproud StokeonTrent Live – we’re #localandproudStokeonTrent Live – we’re #localandproudVideo Player is loading.Play VideoPauseUnmute0:02/1:20Loaded: 0%0:02Progress: 0%Stream TypeLIVE-1:18 SharePlayback Rate1xChaptersChaptersDescriptionsdescriptions off, selectedSubtitlessubtitles settings, opens subtitles settings dialogsubtitles off, selectedAudio Trackdefault, selectedFullscreenThis is a modal window.Beginning of dialog window. Escape will cancel and close the window.TextColorWhiteBlackRedGreenBlueYellowMagentaCyanTransparencyOpaqueSemi-TransparentBackgroundColorBlackWhiteRedGreenBlueYellowMagentaCyanTransparencyOpaqueSemi-TransparentTransparentWindowColorBlackWhiteRedGreenBlueYellowMagentaCyanTransparencyTransparentSemi-TransparentOpaqueFont Size50%75%100%125%150%175%200%300%400%Text Edge StyleNoneRaisedDepressedUniformDropshadowFont FamilyProportional Sans-SerifMonospace Sans-SerifProportional SerifMonospace SerifCasualScriptSmall CapsReset restore all settings to the default valuesDoneClose Modal DialogEnd of dialog window.Close Modal DialogThis is a modal window. This modal can be closed by pressing the Escape key or activating the close button.PlayMute0:00/0:00Loaded: 0%Progress: 0%Stream TypeLIVE0:00 Playback Rate1xFullscreenClose Modal DialogThis is a modal window. This modal can be closed by pressing the Escape key or activating the close button. UP NEXT:UP NEXT: Click for Sound Watch again Video will play in Video Loading Video Unavailable We pay for stories! Send your videos to email@example.comWelcome to The Sentinel’s breaking news service bringing you all the latest updates from Stoke-on-Trent and North Staffordshire on Thursday, June 7. Our team of reporters will be updating this live service with all the latest on the weather, traffic and travel as well as news, sport and entertainment through the day. We’ll be bringing you the very latest updates in our live news feed below. For the latest news and breaking news visit www.stokeontrentlive.co.uk Get all the big headlines, pictures, analysis, opinion and video on the stories that matter to you. Follow us on Twitter @SOTLive – the official Sentinel account – real news in real time. We’re also on Facebook – your must-see news, features, videos and pictures throughout Stoke-on-Trent, North Staffordshire & South Cheshire. You’ll also find us on Instagram here . 23:02That’s all for todayJoin us tomorrow from 7am for the latest news, weather, travel updates and more.22:55Weather forecast for tonightAccording to the Met Office, cloud will thicken overnight bringing the risk of some showery bursts of rain, perhaps heavy and thundery in places. It will be warm and humid for all. Minimum temperature 12 °C.20:46Accident in Dresden now clearedThe accident on Carlisle Street in Dresden has now cleared and traffic is now moving in both directions between Peel Street and A5035 Belgrave Road, near Bargain Booze. 19:30This ‘cheese guru’ from Trentham Shopping Village is one of the country’s best!Nicola Beardmore proved she was ‘whey’ out in front of her rivals – after winning the Young Cheesemonger of the Year award.Nicola overcame seven other competitors from across the country after being invited to contest the prize at the British Cheese Awards in Somerset.Nicola excelled in rounds such as a cheeseboard discussion, a cut and wrap exercise, identification test and ‘Masterrind’ – a Mastermind-style quiz – in the final at the Royal Bath and West Show.She had earlier won through to the big event by writing a brief history of her work and choosing how she would spend £20 on cheese for six people.Nicola, who works at the Brown and Green Deli at the Trentham Estate, has now returned home to Longton with her prizes – a trophy and ‘cheese iron’ which is used to take samples.Read the full report here.Award-winning Nicola Beardmore with her trophies19:07Caravan ‘towed’ before travellers leave popular city parkTravellers have been evicted from a park – with one caravan even spotted being towed away.A group had set up camp at Fenton’s Smithpool Park on Tuesday night with 12 vehicles including vans and caravans.Read the full report here.A caravan was seen on the back of a recovery truck (Pic: Jessica Stone) (Image: Jessica Stone)18:57Dresden street blocked after accidentCarlisle Street in Dresden is blocked in both directions due to an accident between Peel Street and A5035 Belgrave Road, near Bargain Booze.18:40Where to see the cutest baby zoo animals born around Stoke-on-TrentIt’s officially the best time of year to visit the zoo. Why? Because there’s cute baby animals in abundance right now. With little penguins and elephants, plus many more breeds, hatching over the past few weeks, the whole family will be in awe at all the adorableness going on.See them all here!One of the chicks18:16‘Michael Jackson’ has lucky escape after getting stuck in trailerA horse called Michael Jackson had a lucky escape this afternoon – after getting stuck in a trailer.Fire crews from Cannock responded to a call about the 10-year-old cob, who was in trouble near Hatherton.Michael Jackson was an eighteenth birthday present, and had just been collected when he got his front legs caught over the breast bar of the trailer. Read our full story here. Firefighters working to release Michael Jackson the horse. (Image: Staffordshire Fire and Rescue)17:53‘Serious fire’ at terraced house sparked after dog knocks unattended candle overAn investigation into a ‘serious house fire’ in Newcastle this afternoon has revealed the blaze was caused by a dog knocking an unattended candle over.Fire crews from Newcastle, Hanley and Sandyford were called to Orme Roadjust after 11.15am.The blaze started in a bedroom and two adults and two children managed to get out of the terraced house unharmed, although one of them was checked over by paramedics for smoke inhalation.Read the full story here.The window of the bedroom where the fire was sparked (Image: Staffordshire Fire and Rescue)16:30Stop-start traffic on M6 SouthboundWe have reports of one lane blocked, with stop-start traffic, due to an accident on the M6 Southbound between J19 A556 (Knutsford) and J18 A54 (Middlewich / Holmes Chapel). Travel time is 35 minutes, and was already heavy in the area following an earlier breakdown. 16:24Traffic hotspots- There is slow traffic on the A500 D Road Northbound between A50 (Sideway Roundabout) and A527 Grange Road (Wolstanton Retail Park).- Slow traffic on the A50 Market Place in both directions between B5051 Moorland Road and A53 Elder Road (Cobridge Traffic Lights). – Usual slow traffic on the A534 Crewe Green Road Eastbound before the roadworks at A5020 University Way (Crewe Green roundabout). 16:12Accident on the motorwayOne lane of the M6 is closed and there is queueing traffic due to an accident on Northbound between Keele Services and J16 A500 (Crewe / Stoke-On-Trent).Lane three (of three) is closed.16:06Slow trafficThere is slow traffic and traffic is heavier than normal on the M6 Northbound between Keele Services and J16 A500 (Crewe / Stoke-On-Trent).14:51Delays on the motorway and A500There are currently delays on the M6 southbound near to Sandbach and slow traffic on the A500 northbound between Alsager Road (Audley turn off) and M6 J16 (Crewe / Stoke-On-Trent). if you are heading out that way you may want to give yourself some extra time. 13:58Two lanes of the M6 closedThere is queueing traffic and two lanes of the M6 are closed due to a broken down lorry on Southbound at J17 A534 (Sandbach / Crewe). In the roadworks area. Lanes one and two (of three) are now closed opposite the entry slip road. This is breakdown has occurred within the residual delays of an earlier broken down van. 13:28M6 southbound in CheshireSlow traffic and one lane is blocked due to a broken down van on the M6 southbound between junction 17, Sandbach, and junction 16, Crewe. Lane one, of three, is blocked inside the smart motorway area.13:25Police issue warning about rogue tradersPolice officers are warning families to be vigilant about rogue traders and unsolicited cold callers.Detectives are warning people, especially the elderly, to be on their guard having received reports of rogue traders in the Weston Coyney area.The cold callers often say they are from an official sounding organisation such as the ‘water board’ or offer to complete work the homeowners may not want.Read more.13:18A500 D-roadQueuing traffic on the A500 D Road heading southbound between Stone Road and the M6.12:59M6 northboundOne lane is closed and traffic is queuing because of a broken down lorry on the M6 northbound between junction 16, Crewe, and junction 17, Sandbach, in the roadworks area. Lane one, of three, is closed and recovery work is on going. 12:40Firefighters called to ‘serious house fire’ in terrace street (Image: Staffordshire Fire and Rescue Service)Firefighters rushed to a terrace house after receiving reports of a ‘serious house fire’.Appliances from Newcastle, Hanley, and Sandyford were called to Orme Road, Newcastle at around 11.15am today (June 7) to reports of the fire.When they arrived crews found the property was empty and the cause of the fire is not yet known.Read more.10:21Find out who is in court todayToday’s cases include a 40-year-old man, from Penkhull, accused of using a stolen credit card, a 69-year-old Chell woman charged with drink driving and a 22-year-old man accused of harassing a Hanley woman.Follow all the latest court action here. 09:43Firefighters reveal what caused major skip yard fireFirefighters believe arsonists could have targeted a skip yard last night after plumes of smoke were spotted across North Staffordshire.Fire crews from Sandyford and Hanley along with a water carrier from Cheadle were at a skip yard just off Reginald Mitchell Way in Sandyford for more than three hours as they tackled the blaze.Read more.08:53Crewe congestionCongestion on University Way in Crewe heading towards the Crewe Green Roundabout.08:20A34, Congleton Heavy traffic on A34 Manchester Road in both directions between A536 and Moss Lane. Temporary traffic lights are in operation. 07:52A34 Newcastle Road Slow traffic on A34 Newcastle Road Northbound between Child’s Lane and Watery Lane. In the roadworks area. Temporary traffic lights are in operation. 07:47KingsleyReports of accident on A521 Froghall Road at Hammersley Hayes Road. Traffic is coping well. 07:24Find out why this stolen car was travelling into Staffordshire (Image: Staffs Police TST)A man has been arrested after being pulled over by police officers while travelling in a stolen car. The Staffordshire Police Tactical Support Team (TST) stopped the Mercedes this morning as it was travelling into the county from the West Midlands. Officers attempted to stop the vehicle earlier today (June 7) but the driver abandoned the car and tried to make a getaway.Read more.06:58A50 UttoxeterReports of traffic problem and queuing traffic on A50 Eastbound at B5030 Ashbourne Road.Weather forecastToday is set to be a sunny day as the Met Office predicts temperatures could hit around 21 °C. After some early patchy cloud, some good sunshine will develop this morning into the afternoon. Cloud will start to build by the mid afternoon however, with areas of rain perhaps arriving in the south later, bringing the risk of thunder. The band of potentially thundery rain will affect some places through the night, slowly moving northwards. Outside of the rain skies will stay cloudy too. Good morningGood morning and welcome to the StokeonTrentLive news feed for the day. We will be bringing you traffic updates as well as the latest breaking news from Stoke-on-Trent, Staffordshire, and South Cheshire. Share this video Watch Next Click to playTap to play
With temperatures in the low to mid-40s expected once again today, air-conditioners, refrigeration plants and electricity use could again pose a danger to the town’s electricity supply.Two town-wide outages took place yesterday (Tuesday) as electricity use sky-rocketed, with air-conditioners and refrigeration plants working hard to keep things cool.The demand was just too much, causing a system-wide emergency shutdown.Power was restored twice before officials appealed to residents to limit use of air-conditioners and other electricity-intensive appliances, if possible.This seemed to have the desired result, with electricity supply remaining stable.Today, expect sporadic outages to continue to occur as the system strains to meet demand.Everyone is asked to turn off air-conditioners that are not needed. WebsiteWebsiteWebsite WebsiteWebsiteWebsite WebsiteWebsiteWebsite
GANs are neural networks used in unsupervised learning that generate synthetic data given certain input data. GAN’s have two components: a generator and a discriminator. A generator generates new instances of an object and the discriminator determines whether the new instance belongs to the actual dataset. A generative learn how the data is generated i.e. the structure of the data, in order to categorize it. This allows the system to generate samples with similar statistical properties. Discriminative models will learn the relation between the data and the label associated with the data. The discriminative model will categorize the input data without knowing how the data is generated. GAN exploits the concept behind both the models to get a better network architecture. This tutorial on GAN’s will help you build a neural network that fills in the missing part of a handwritten digit. This tutorial will cover how to build an MNIST digit classifier and simulate a dataset of handwritten digits with sections of the handwritten numbers missing. Next, users will learn using the MNIST classifier to predict on noised/masked MNIST digits dataset (simulated dataset) and implement GAN to generate back the missing regions of the digit. This tutorial will also cover using the MNIST classifier to predict on the generated digits from GAN and finally compare performance between masked data and generated data. This tutorial is an excerpt from a book written by Matthew Lamons, Rahul Kumar, Abhishek Nagaraja titled Python Deep Learning Projects. This book will help users develop their own deep learning systems in a straightforward way and in an efficient way. The book has projects developed using complex deep learning projects in the field of computational linguistics and computer vision to help users master the subject. All of the Python files and Jupyter Notebook files for this tutorial can be found at GitHub. In this tutorial, we will be using the Keras deep learning library. Importing all of the dependencies We will be using numpy, matplotlib, keras, tensorflow, and the tqdm package in this exercise. Here, TensorFlow is used as the backend for Keras. You can install these packages with pip. For the MNIST data, we will be using the dataset available in the keras module with a simple import: import numpy as npimport randomimport matplotlib.pyplot as plt%matplotlib inline from tqdm import tqdmfrom keras.layers import Input, Conv2Dfrom keras.layers import AveragePooling2D, BatchNormalizationfrom keras.layers import UpSampling2D, Flatten, Activationfrom keras.models import Model, Sequentialfrom keras.layers.core import Dense, Dropoutfrom keras.layers.advanced_activations import LeakyReLUfrom keras.optimizers import Adamfrom keras import backend as kfrom keras.datasets import mnist It is important that you set seed for reproducibility: # set seed for reproducibilityseed_val = 9000np.random.seed(seed_val)random.seed(seed_val) Exploring the data We will load the MNIST data into our session from the keras module with mnist.load_data(). After doing so, we will print the shape and the size of the dataset, as well as the number of classes and unique labels in the dataset: (X_train, y_train), (X_test, y_test) = mnist.load_data() print(‘Size of the training_set: ‘, X_train.shape)print(‘Size of the test_set: ‘, X_test.shape)print(‘Shape of each image: ‘, X_train.shape)print(‘Total number of classes: ‘, len(np.unique(y_train)))print(‘Unique class labels: ‘, np.unique(y_train)) We have a dataset with 10 different classes and 60,000 images, with each image having a shape of 28*28 and each class having 6,000 images. Let’s plot and see what the handwritten images look like: # Plot of 9 random imagesfor i in range(0, 9): plt.subplot(331+i) # plot of 3 rows and 3 columns plt.axis(‘off’) # turn off axis plt.imshow(X_train[i], cmap=’gray’) # gray scale The output is as follows: Let’s plot a handwritten digit from each class: # plotting image from each classfig=plt.figure(figsize=(8, 4))columns = 5rows = 2for i in range(0, rows*columns): fig.add_subplot(rows, columns, i+1) plt.title(str(i)) # label plt.axis(‘off’) # turn off axis plt.imshow(X_train[np.where(y_train==i)], cmap=’gray’) # gray scaleplt.show() The output is as follows: Look at the maximum and the minimum pixel value in the dataset: print(‘Maximum pixel value in the training_set: ‘, np.max(X_train))print(‘Minimum pixel value in the training_set: ‘, np.min(X_train)) The output is as follows: Preparing the data Type conversion, centering, scaling, and reshaping are some of the pre-processing we will implement in this tutorial. Type conversion, centering and scaling Set the type to np.float32. For centering, we subtract the dataset by 127.5. The values in the dataset will now range between -127.5 to 127.5. For scaling, we divide the centered dataset by half of the maximum pixel value in the dataset, that is, 255/2. This will result in a dataset with values ranging between -1 and 1: # Converting integer values to float types X_train = X_train.astype(np.float32)X_test = X_test.astype(np.float32) # Scaling and centeringX_train = (X_train – 127.5) / 127.5X_test = (X_test – 127.5)/ 127.5print(‘Maximum pixel value in the training_set after Centering and Scaling: ‘, np.max(X_train))print(‘Minimum pixel value in the training_set after Centering and Scaling: ‘, np.min(X_train)) Let’s define a function to rescale the pixel values of the scaled image to range between 0 and 255: # Rescale the pixel values (0 and 255)def upscale(image): return (image*127.5 + 127.5).astype(np.uint8) # Lets see if this worksz = upscale(X_train)print(‘Maximum pixel value after upscaling scaled image: ‘,np.max(z))print(‘Maximum pixel value after upscaling scaled image: ‘,np.min(z)) A plot of 9 centered and scaled images after upscaling: for i in range(0, 9): plt.subplot(331+i) # plot of 3 rows and 3 columns plt.axis(‘off’) # turn off axis plt.imshow(upscale(X_train[i]), cmap=’gray’) # gray scale The output is as follows: Masking/inserting noise For the needs of this project, we need to simulate a dataset of incomplete digits. So, let’s write a function to mask small regions in the original image to form the noised dataset. The idea is to mask an 8*8 region of the image with the top-left corner of the mask falling between the 9th and 13th pixel (between index 8 and 12) along both the x and y axis of the image. This is to make sure that we are always masking around the center part of the image: def noising(image): array = np.array(image) i = random.choice(range(8,12)) # x coordinate for the top left corner of the mask j = random.choice(range(8,12)) # y coordinate for the top left corner of the mask array[i:i+8, j:j+8]=-1.0 # setting the pixels in the masked region to -1 return array noised_train_data = np.array([*map(noising, X_train)])noised_test_data = np.array([*map(noising, X_test)])print(‘Noised train data Shape/Dimension : ‘, noised_train_data.shape)print(‘Noised test data Shape/Dimension : ‘, noised_train_data.shape) A plot of 9 scaled noised images after upscaling: # Plot of 9 scaled noised images after upscalingfor i in range(0, 9): plt.subplot(331+i) # plot of 3 rows and 3 columns plt.axis(‘off’) # turn off axis plt.imshow(upscale(noised_train_data[i]), cmap=’gray’) # gray scale The output is as follows: Reshaping Reshape the original dataset and the noised dataset to a shape of 60000*28*28*1. This is important since the 2D convolutions expect to receive images of a shape of 28*28*1: # Reshaping the training dataX_train = X_train.reshape(X_train.shape, X_train.shape, X_train.shape, 1)print(‘Size/Shape of the original training set: ‘, X_train.shape) # Reshaping the noised training datanoised_train_data = noised_train_data.reshape(noised_train_data.shape,noised_train_data.shape,noised_train_data.shape, 1)print(‘Size/Shape of the noised training set: ‘, noised_train_data.shape)# Reshaping the testing dataX_test = X_test.reshape(X_test.shape, X_test.shape, X_test.shape, 1)print(‘Size/Shape of the original test set: ‘, X_test.shape)# Reshaping the noised testing datanoised_test_data = noised_test_data.reshape(noised_test_data.shape,noised_test_data.shape,noised_test_data.shape, 1)print(‘Size/Shape of the noised test set: ‘, noised_test_data.shape) MNIST classifier To start off with modeling, let’s build a simple convolutional neural network (CNN) digit classifier. The first layer is a convolution layer that has 32 filters of a shape of 3*3, with relu activation and Dropout as the regularizer. The second layer is a convolution layer that has 64 filters of a shape of 3*3, with relu activation and Dropout as the regularizer. The third layer is a convolution layer that has 128 filters of a shape of 3*3, with relu activation and Dropout as the regularizer, which is finally flattened. The fourth layer is a Dense layer of 1024 neurons with relu activation. The final layer is a Dense layer with 10 neurons corresponding to the 10 classes in the MNIST dataset, and the activation used here is softmax, batch_size is set to 128, the optimizer used is adam, and validation_split is set to 0.2. This means that 20% of the training set will be used as the validation set: # input image shapeinput_shape = (28,28,1) def train_mnist(input_shape, X_train, y_train):model = Sequential()model.add(Conv2D(32, (3, 3), strides=2, padding=’same’,input_shape=input_shape))model.add(Activation(‘relu’))model.add(Dropout(0.2))model.add(Conv2D(64, (3, 3), strides=2, padding=’same’))model.add(Activation(‘relu’))model.add(Dropout(0.2))model.add(Conv2D(128, (3, 3), padding=’same’))model.add(Activation(‘relu’))model.add(Dropout(0.2))model.add(Flatten())model.add(Dense(1024, activation = ‘relu’))model.add(Dense(10, activation=’softmax’))model.compile(loss = ‘sparse_categorical_crossentropy’,optimizer = ‘adam’, metrics = [‘accuracy’])model.fit(X_train, y_train, batch_size = 128, epochs = 3, validation_split=0.2, verbose = 1 )return modelmnist_model = train_mnist(input_shape, X_train, y_train) The output is as follows: Use the built CNN digit classifier on the masked images to get a measure of its performance on digits that are missing small sections: # prediction on the masked imagespred_labels = mnist_model.predict_classes(noised_test_data)print(‘The model model accuracy on the masked images is:’,np.mean(pred_labels==y_test)*100) On the masked images, the CNN digit classifier is 74.9% accurate. It might be slightly different when you run it, but it will still be very close. Defining hyperparameters for GAN The following are some of the hyperparameters defined that we will be using throughout the code and are totally configurable: # Smoothing valuesmooth_real = 0.9 # Number of epochsepochs = 5# Batchsizebatch_size = 128# Optimizer for the generatoroptimizer_g = Adam(lr=0.0002, beta_1=0.5)# Optimizer for the discriminatoroptimizer_d = Adam(lr=0.0004, beta_1=0.5)# Shape of the input imageinput_shape = (28,28,1) Building the GAN model components With the idea that the final GAN model will be able to fill in the part of the image that is missing (masked), let’s define the generator. You can understand how to define the generator, discriminator, and DCGAN by referring to our book. Training GAN We’ve built the components of the GAN. Let’s train the model in the next steps! Plotting the training – part 1 During each epoch, the following function plots 9 generated images. For comparison, it will also plot the corresponding 9 original target images and 9 noised input images. We need to use the upscale function we’ve defined when plotting to make sure the images are scaled to range between 0 and 255, so that you do not encounter issues when plotting: def generated_images_plot(original, noised_data, generator): print(‘NOISED’)for i in range(9):plt.subplot(331 + i)plt.axis(‘off’)plt.imshow(upscale(np.squeeze(noised_data[i])), cmap=’gray’) # upscale for plottingplt.show()print(‘GENERATED’)for i in range(9):pred = generator.predict(noised_data[i:i+1], verbose=0)plt.subplot(331 + i)plt.axis(‘off’)plt.imshow(upscale(np.squeeze(pred)), cmap=’gray’) # upscale to avoid plotting errorsplt.show()print(‘ORIGINAL’)for i in range(9):plt.subplot(331 + i)plt.axis(‘off’)plt.imshow(upscale(np.squeeze(original[i])), cmap=’gray’) # upscale for plottingplt.show() The output of this function is as follows: Plotting the training – part 2 Let’s define another function that plots the images generated during each epoch. To reflect the difference, we will also include the original and the masked/noised images in the plot. The top row contains the original images, the middle row contains the masked images, and the bottom row contains the generated images. The plot has 12 rows with the sequence, row 1 – original, row 2 – masked, row3 – generated, row 4 – original, row5 – masked,…, row 12 – generated. Let’s take a look at the code for the same: def plot_generated_images_combined(original, noised_data, generator): rows, cols = 4, 12 num = rows * cols image_size = 28 generated_images = generator.predict(noised_data[0:num])imgs = np.concatenate([original[0:num], noised_data[0:num], generated_images])imgs = imgs.reshape((rows * 3, cols, image_size, image_size))imgs = np.vstack(np.split(imgs, rows, axis=1))imgs = imgs.reshape((rows * 3, -1, image_size, image_size))imgs = np.vstack([np.hstack(i) for i in imgs])imgs = upscale(imgs)plt.figure(figsize=(8,16))plt.axis(‘off’)plt.title(‘Original Images: top rows, ”Corrupted Input: middle rows, ”Generated Images: bottom rows’)plt.imshow(imgs, cmap=’gray’)plt.show() The output is as follows: Training loop Now we are at the most important part of the code; the part where all of the functions we previously defined will be used. The following are the steps: Load the generator by calling the img_generator() function. Load the discriminator by calling the img_discriminator() function and compile it with the binary cross-entropy loss and optimizer as optimizer_d, which we have defined under the hyperparameters section. Feed the generator and the discriminator to the dcgan() function and compile it with the binary cross-entropy loss and optimizer as optimizer_g, which we have defined under the hyperparameters section. Create a new batch of original images and masked images. Generate new fake images by feeding the batch of masked images to the generator. Concatenate the original and generated images so that the first 128 images are all original and the next 128 images are all fake. It is important that you do not shuffle the data here, otherwise it will be hard to train. Label the generated images as 0 and original images as 0.9 instead of 1. This is one-sided label smoothing on the original images. The reason for using label smoothing is to make the network resilient to adversarial examples. It’s called one-sided because we are smoothing labels only for the real images. Set discriminator.trainable to True to enable training of the discriminator and feed this set of 256 images and their corresponding labels to the discriminator for classification. Now, set discriminator.trainable to False and feed a new batch of 128 masked images labeled as 1 to the GAN (DCGAN) for classification. It is important to set discriminator.trainable to False to make sure the discriminator is not getting trained while training the generator. Repeat steps 4 through 7 for the desired number of epochs. We have placed the plot_generated_images_combined() function and the generated_images_plot() function to get a plot generated by both functions after the first iteration in the first epoch and after the end of each epoch. Feel free to place these plot functions according to the frequency of plots you need displayed: def train(X_train, noised_train_data, input_shape, smooth_real, epochs, batch_size, optimizer_g, optimizer_d): # define two empty lists to store the discriminator # and the generator lossesdiscriminator_losses = generator_losses = # Number of iteration possible with batches of size 128iterations = X_train.shape // batch_size# Load the generator and the discriminatorgenerator = img_generator(input_shape)discriminator = img_discriminator(input_shape)# Compile the discriminator with binary_crossentropy lossdiscriminator.compile(loss=’binary_crossentropy’,optimizer=optimizer_d)# Feed the generator and the discriminator to the function dcgan # to form the DCGAN architecturegan = dcgan(discriminator, generator, input_shape)# Compile the DCGAN with binary_crossentropy lossgan.compile(loss=’binary_crossentropy’, optimizer=optimizer_g)for i in range(epochs):print (‘Epoch %d’ % (i+1))# Use tqdm to get an estimate of time remainingfor j in tqdm(range(1, iterations+1)):# batch of original images (batch = batchsize)original = X_train[np.random.randint(0, X_train.shape, size=batch_size)]# batch of noised images (batch = batchsize)noise = noised_train_data[np.random.randint(0, noised_train_data.shape, size=batch_size)]# Generate fake imagesgenerated_images = generator.predict(noise)# Labels for generated datadis_lab = np.zeros(2*batch_size)# data for discriminatordis_train = np.concatenate([original, generated_images])# label smoothing for original imagesdis_lab[:batch_size] = smooth_real# Train discriminator on original imagesdiscriminator.trainable = Truediscriminator_loss = discriminator.train_on_batch(dis_train, dis_lab)# save the losses discriminator_losses.append(discriminator_loss)# Train generatorgen_lab = np.ones(batch_size)discriminator.trainable = Falsesample_indices = np.random.randint(0, X_train.shape, size=batch_size)original = X_train[sample_indices]noise = noised_train_data[sample_indices]generator_loss = gan.train_on_batch(noise, gen_lab)# save the lossesgenerator_losses.append(generator_loss)if i == 0 and j == 1:print(‘Iteration – %d’, j)generated_images_plot(original, noise, generator)plot_generated_images_combined(original, noise, generator)print(“Discriminator Loss: “, discriminator_loss,\”, Adversarial Loss: “, generator_loss)# training plot 1generated_images_plot(original, noise, generator)# training plot 2plot_generated_images_combined(original, noise, generator)# plot the training lossesplt.figure()plt.plot(range(len(discriminator_losses)), discriminator_losses,color=’red’, label=’Discriminator loss’)plt.plot(range(len(generator_losses)), generator_losses,color=’blue’, label=’Adversarial loss’)plt.title(‘Discriminator and Adversarial loss’)plt.xlabel(‘Iterations’)plt.ylabel(‘Loss (Adversarial/Discriminator)’)plt.legend()plt.show()return generatorgenerator = train(X_train, noised_train_data,input_shape, smooth_real,epochs, batch_size,optimizer_g, optimizer_d) The output is as follows: Generated images plotted with training plots at the end of the first iteration of epoch 1 Generated images plotted with training plots at the end of epoch 2 Generated images plotted with training plots at the end of epoch 5 Plot of the discriminator and adversarial loss during training Predictions CNN classifier predictions on the noised and generated images We will call the generator on the masked MNIST test data to generate images, that is, fill in the missing part of the digits: # restore missing parts of the digit with the generatorgen_imgs_test = generator.predict(noised_test_data) Then, we will pass the generated MNIST digits to the digit classifier we have modeled already: # predict on the restored/generated digitsgen_pred_lab = mnist_model.predict_classes(gen_imgs_test)print(‘The model model accuracy on the generated images is:’,np.mean(gen_pred_lab==y_test)*100) The MNIST CNN classifier is 87.82% accurate on the generated data. The following is a plot showing 10 generated images by the generator, the actual label of the generated image, and the label predicted by the digit classifier after processing the generated image: # plot of 10 generated images and their predicted labelfig=plt.figure(figsize=(8, 4))plt.title(‘Generated Images’)plt.axis(‘off’) columns = 5rows = 2for i in range(0, rows*columns): fig.add_subplot(rows, columns, i+1) plt.title(‘Act: %d, Pred: %d’%(gen_pred_lab[i],y_test[i])) # label plt.axis(‘off’) # turn off axis plt.imshow(upscale(np.squeeze(gen_imgs_test[i])), cmap=’gray’) # gray scaleplt.show() The output is as follows: The Jupyter Notebook code files for the preceding DCGAN MNIST inpainting can be found at GitHub. Use the Jupyter Notebook code files for the DCGAN Fashion MNIST inpainting can be found. Summary We built a deep convolution GAN in Keras on handwritten MNIST digits and understood the function of the generator and the discriminator component of the GAN. We defined key hyperparameters, as well as, in some places, reasoned with why we used what we did. Finally, we tested the GAN’s performance on unseen data and determined that we succeeded in achieving our goals. To understand insightful projects to master deep learning and neural network architectures using Python and Keras, check out this book Python Deep Learning Projects. Read Next Getting started with Web Scraping using Python [Tutorial] Google researchers introduce JAX: A TensorFlow-like framework for generating high-performance code from Python and NumPy machine learning programs Google releases Magenta studio beta, an open source python machine learning library for music artists
Posted by << Previous PostNext Post >> Tags: Norwegian Cruise Line NCL’s January promotions include free onboard spending, complimentary amenities & more MIAMI — Norwegian Cruise Line has announced a Super Splash Sale to kick off the New Year, giving passengers up to US$500 in free onboard spending on select five-day or longer sailings.Now through Jan. 15, the offer is also combinable with Free at Sea, providing up to $3,100 in added value to a client’s cruise.Free at Sea has been extended through Jan. 31 and offers guests who book a new three-day or longer sailing in an oceanview, balcony or minisuite category stateroom with the chance to select two free choices from five onboard amenities. These include free unlimited open bar, a free specialty dining package, $50 towards shore excursions in each port per stateroom, 250 minutes of free Wi-Fi, or the Friends & Family Sail Free offer that allows the third and fourth guests to sail for free on many 2018 sail dates. Inside staterooms, as well as Norwegian’s Studio staterooms for solo travellers, can choose one free offer. Those in a suite of The Haven by Norwegian enjoy all five free amenities.More news: Canada raises travel warning amid escalating protests in Hong KongOther January offers include Free at Sea Hawaii, which has also been extended through Jan. 31. Guests who book a new cruise aboard Pride of America – the only ship in the world that can sail seven-day inter-island cruises in Hawaii – can select one of five offers. These include a free one-night free-cruise hotel stay with an option to select reduced airfare starting at $399 from select gateways, a free specialty dining package, free pre-paid service charges for the first and second guest in the stateroom, free $50 per port shore excursion credit per stateroom, or Friends & Family Sail Free. Guests who book a suite can enjoy all five offers.Clients who wish to sail to Cuba this year can also take advantage of $50 in free onboard spending money with Norwegian’s Havana Nights promotion, running now through Jan. 31. Those who book a new cruise to Cuba can enjoy an all-inclusive onboard experience with unlimited open bar on Norwegian Sky from Miami or Norwegian Sun from Port Canaveral. Share Travelweek Group Wednesday, January 3, 2018
TEMPE, Ariz. — If once is a fluke, twice is a trend and three times is a habit, what does six times make?That’s the question the Arizona Cardinals are trying to answer on a short week following another lackluster defensive performance against an opposing team’s tight end.Sunday’s culprit in the team’s 32-20 loss to the San Francisco 49ers was Pro Bowler Vernon Davis.Davis, who came into the contest with 14 catches for 244 yards on the season, torched Todd Bowles’ unit — in particular the secondary — en route to a career day. The former first-round pick caught seven balls in the first half for 171 yards — the most receiving yards amassed by any player in a half this season. In fact, it marked the first time a tight end reached the 171-yard mark in a half since Hall of Famer Shannon Sharpe recorded the feat back in Oct. 2002. Comments Share Grace expects Greinke trade to have emotional impact Derrick Hall satisfied with D-backs’ buying and selling Former Cardinals kicker Phil Dawson retires Top Stories Head coach Bruce Arians, however, believes it’s a sign of the times in the league and not a problem exclusive to Arizona’s defense.“I think that’s a pattern throughout the NFL,” Arians said. “It’s not just us. The tight ends are becoming more athletic and more teams are targeting them.”Maybe so, but the fact of the matter is Cook and Davis will each make another appearance at University of Phoenix Stadium this year, as will Atlanta’s future Hall-of-Famer Tony Gonzalez.So how do the Cardinals plan to crack the seemingly unbreakable code?“We’ve just got to start recognizing [opposing tight ends’] strong points,” Powers said. “Obviously, Vernon was their go-to guy along with Anquan [Boldin], and we didn’t do a good job stopping them. That’s definitely going to be a key factor for us down the stretch, figuring out how to stop tight ends and whoever else their weapons are.”Powers noted that it’s as much about a mindset as anything else when it comes to game planning against tight ends.“I think we’ve got to start taking it more seriously,” said Powers. “Sometimes when we line up on tight ends we start to think that it is what is.” The 5: Takeaways from the Coyotes’ introduction of Alex Meruelo Of those seven first half catches, two of them went for touchdowns of 35 and 61 yards. On the first score, Davis breezed into the second level of the Cardinals’ secondary and was never touched by safety Yeremiah Bell or cornerback Jerraud Powers. On the second score, the 6-foot-3 Davis simply outran and out-jumped Bell.“Vernon’s a tough matchup,” Bell said. “There’s no doubt about it. There are good tight ends around the league and they’re tough matchups. But like every week, you win some you lose.”The problem for Arizona so far in 2013 is that every week they seem to be losing the matchup against tight ends, turning Davis, Jared Cook and Jimmy Graham into the next three Canton-bound candidates to join Sharpe.Week 1: Jared Cook – seven catches, 141 yards and two touchdownsWeek 2: Brandon Pettigrew – three catches, 32 yardsWeek 3: Jimmy Graham – nine catches, 134 yards and two touchdownsWeek 4: Timothy Wright – five catches for 41 yards Week 5: Greg Olsen – five catches, 79 yardsWeek 6: Vernon Davis – eight catches, 180 yards and two touchdownsThe tight end output against the Cardinals this season hasn’t been prolific every week, but it’s been a consistent theme opposing teams have tried to exploit in the passing game. Bell was in agreement with Powers on that front and added that collectively the unit ‘has to play sound defense.’“Your eyes have to be in the right place at the right time snap after snap. And you just have to stay focused.”Through six games, Arizona has allowed starting tight ends to catch 37 passes for 607 yards and six touchdowns. Thursday’s trio – Seattle’s Zach Miller, Luke Willson and Kellen Davis — has 18 receptions for 228 yards and two touchdowns on the season. – / 22