AI Weekly Malaysia

Items

Browse the latest source items collected for AI Weekly Malaysia. Open any item to see the original source, context, and related AI-generated summary when available.

Reset

Showing 6951-6975 of 7010 results

DateProviderCategoryStatusItem
16 Aug 2017, 3:00 PMOpenAI Newsai-labsnew item More on Dota 2

Our Dota 2 result shows that self-play can catapult the performance of machine learning systems from far below human level to superhuman, given sufficient compute. In the span of a month, our system went from barely matching a high-ranked player to beating the top pros and has continued to improve since then. Supervised deep learning systems can only be as good as their training datasets, but in self-play systems, the available data improves automatically as the agent gets better.

11 Aug 2017, 10:47 PMDigital News Asiamalaysia-technew item Mobile game Vainglory raises US$19 mil in additional funding

Latest round of funding to be used to expand development team and improve game engine Vainglory to make a big push in SEA market with more activities in store Mobile games maker Super Evil Megacorp, creators of Vainglory a game for iOS and Android devices, has raised US$19 million (RM81.5 million) in funding from its existing investors which include Index Ventures, Korea Investment Partners, Signia Venture Partners, General Catalyst among several others. Vainglory is a mobile game of the multiplayer online battle arena (MOBA) genre that is similar to games like Dota 2 and League of Legends. Super Evil Megacorp will utilise the financing round to expand its team and deliver on its mission to build the best MOBA experience for touchscreen gamers. “Already the most competitive mobile MOBA on the market, with an active global esports scene and growing community, Vainglory is off to a great start,” Super Evil Megacorp chief executive officer Kristian Segerstrale said. “Together with our investors we’re doubling down on our missions to build the best core gaming experiences for the touchscreen generation, and to create a home for talented game developers who are passionate about pushing the boundaries of the industry,” he said. Segerstrale explained that the funding will be utilised to grow the company’s team of game designers, engineers and artists. Super Evil Megacorp will also use the investment to continue developing its proprietary multi-platform E.V.I.L. engine that has allowed Vainglory to deliver stunning graphics and beautiful visuals. Currently, the studio is heavily focused on developing Vainglory’s upcoming 5v5 game mode, which will add to the strategic depth of the game and offer more to the already thriving Vainglory competitive community. According to the company, Vainglory has made an impact on the mobile game scene just two years after its launch. It claims the game is already the world’s biggest mobile eSports game with more than 3,000 teams across North America and Europe alone, 1,500 teams in APAC and over 1 billion matches played. The company has also partnered with the likes of Apple, Samsung, Google, Amazon, ESL and others to deliver competitions and eSports events for fans across the world. Super Evil Megacorp has recently been making heavy forays into the Southeast Asian (SEA) market having held the SEA Spring Championships in Manila and participated in Singapore’s annual Youth celebrations at the Shine Festival 2017, where they were part of a line-up that featured eSports for the first time in the festival’s 12 year history. The company said it plans to hold the 2017 Vainglory World Championship in Singapore bringing the game full circle as it was tested in early beta in the island republic back in 2013. Related Stories: Netmarble’s Lineage2 Revolution to run first massive fortress siege on July 28 New refreshed Omen gaming PCs from HP New challenges await Lineage 2 Revolution players Keyword(s) : Vaingl

11 Aug 2017, 3:00 PMOpenAI Newsai-labsnew item Dota 2

We’ve created a bot which beats the world’s top professionals at 1v1 matches of Dota 2 under standard tournament rules. The bot learned the game from scratch by self-play, and does not use imitation learning or tree search. This is a step towards building AI systems which accomplish well-defined goals in messy, complicated situations involving real humans.

11 Aug 2017, 1:40 PMDigital News Asiamalaysia-technew item US$7.4 mil funding brings Zanroo closer to unicorn dream

Funding kickstarts high-growth phase and tech expansion Ambitions to be a global top 10 martech company Zanroo founders Chitpol Mungprom (CEO) and Udomsak Donkhampai (CTO) Thai-based insight discovery and marketing technology startup Zanroo is one of the fastest-growing startups in its industry; founders Chitpol Mungprom and Udomsak Donkhampai claim it is the fastest-growing in Southeast Asia (SEA). Zanroo, which was founded in July 2013, helps its customers discover actionable insights through real-time big-data analytics so that these businesses can make their campaign activation and crisis management processes more efficient and successful. Its customer acquisition in the last four years has been significant, certainly helped along by its bespoke proprietary technology. It now has more than 300 leading brands across various verticals, including banking, automotive, telecommunications, e-commerce, aviation and consumer goods. It is active in 15 countries: Malaysia, Singapore, Indonesia, Philippines, Myanmar, Cambodia, Vietnam, Japan, Taiwan, Hong Kong, Australia, Sri Lanka, Pakistan, the UK and Thailand. On July 11, Zanroo closed its Series A funding, securing US$7.4 million (RM31.8 million) from a single investor, Shift Ventures. According to Mungprom, who is chief executive officer, Zanroo will use the funding to propel it further forward in its journey to secure a global footprint and become a unicorn. World domination Zanroo has big plans for global expansion but the founders do not believe they are overly ambitious. The startup is already in 17 countries, its current biggest markets being Mexico and Dubai. The next big markets the startup is looking into are China and cities in Europe, followed by cities in the US. Mungprom says that China will not be too difficult for Zanroo to break into because it already has a local partner that Zanroo obtained through its extensive network in Thailand and because of the good business reputation that the startup has cultivated. Though Mungprom declines to reveal the name of the China partner, he says that it is big in the retail and food and beverage space, and has connections with Alibaba. Zanroo is looking at an expansion timeline of two-and-a-half years, though Mungprom stresses that this timeline is just the first benchmark: “We believe that expansion needs to carry on for the company’s whole life. If you stop expansion the company stops moving,” he says. He elaborates that the funding will be used for two things - hiring people for marketing, sales, and research and development, and to improve on Zanroo’s hardware and software. Zanroo currently has 160 employees across the 17 countries it operates in. “The funding is only for expansion, so we are ready to move up from a startup to a corporation. We want to be the next company that people who come out from Google, Facebook or Agoda come to. We are ready for them,” reveals Mungprom. Though Zanroo is comparatively small, it is moving int

09 Aug 2017, 5:56 PMDigital News Asiamalaysia-technew item Citrix grows leadership team in Asia Pacific Japan

Vincent Wong is named area vice president Citrix Asia Sandeep Pal is regional VP marketing, Orcun Tezel is senior sales director for networking CITRIX on Aug 9 announced that Vincent Wong (pic, above) has been appointed area vice president of Asia for Citrix. Reporting to Stanimira Koleva, vice president of sales and services, Asia Pacific Japan, Citrix, Wong brings extensive experience in driving workspace innovation, cloud adoption and business transformation. He will focus on executing the company’s corporate strategy and delivering consistent growth across all sectors of the Asia region, covering Asean and Korea markets. Wong joins as Citrix is expanding the leadership team across the Asia Pacific Japan region. Citrix recently appointed Sandeep Pal to regional vice president, marketing, Asia Pacific Japan, Citrix and Orcun Tezel as senior sales director for networking, Asia Pacific Japan, Citrix. Commenting on the expanded leadership team, Koleva said: “I’m excited to announce the extended APJ leadership team, and believe Orcun, Sandeep and Vincent will strengthen our ability to support our customers and partners across the region. Our investment and commitment to growth will help drive our business forward and support organisations across the region to transition to cloud and drive digital transformation.” Wong joins Citrix to lead the Asean and Korea sales and services business, bringing almost three decades of experience in business development, product sales, marketing, sales operations, channel management, sales management and business management from companies such as IBM, Cisco Systems and Microsoft. Prior to joining Citrix, Wong spent fourteen years at Microsoft where he held various roles, most recently as director of the company’s APAC channel business. During this time, Wong developed a deep understanding of business transformation and leading change, as well as adapting to the ever-changing needs of the Asia region. “Vincent brings proven expertise and a solid track record to the role. His knowledge of the ASEAN and Korean markets, combined with his strong leadership skills, will strengthen our ability to help businesses in these markets succeed,” said Koleva. Wong holds a BSc degree (with double majors in Computer Science and Information Technology) from The University of Western Australia and an Executive MBA from the California State University, Hayward. Wong was a committee member of the Singapore Chinese Chamber of Commerce and Industry’s IT advisory board, Committee member of the National Infocomm Competency Framework and committee member of Singapore International Chamber of Commerce’s Cybersecurity chapter before he left for Korea in August 2015. He is currently a board member of Netball Singapore, a non-profit sport organisation. Sandeep joins Citrix to lead the Asia Pacific Japan marketing team, with over 25 years of sales and marketing experience in the Asia Pacific region in companies such as IBM and

03 Aug 2017, 3:00 PMOpenAI Newsai-labsnew item Gathering human feedback

RL-Teacher is an open-source implementation of our interface to train AIs via occasional human feedback rather than hand-crafted reward functions. The underlying technique was developed as a step towards safe AI systems, but also applies to reinforcement learning problems with rewards that are hard to specify.

28 Jul 2017, 10:11 AMDigital News Asiamalaysia-technew item HP EliteBook x360:Only the elite need apply

HP's versatile laptop comes with a 360-degree hinge that allows it to transform into one of four modes Beefy security and long last battery provide added value Think of business laptop and you may immediately conjure up an image of a functional but ultimately unexceptionally designed machine that gets the work done but isn’t something you would like to use on a daily basis. That isn’t quite the case with HP’s new EliteBook x360 1030 G2, we will just keep calling it the EliteBook x360 from here on. There is an air of sophistication to this laptop that you don’t see too often in a business laptop. It doesn’t just look good but it also ticks the check boxes in other areas including security, durability and privacy, core features for businesses. Design As far as design goes the EliteBook x360 is a stunning looking business laptop. Coated in silver with an aluminium unibody, it definitely has taken some design cues from Apple’s best. Admittedly, the new stylised HP logo looks strange but apart from that the aesthetics look great. Lifting the lid reveals a 13.3-inch full HD display which produces very vivid colours though it isn’t as bright as other laptops that we have tested in the past. This may make it difficult to read out in bright sunlight but indoor settings, as where most office workers should be, were fine. As it is a convertible laptop, thanks to the 360-degree hinge (hence the x360 monicker) it comes with a handy touchscreen that lets you navigate the EliteBook seamlessly even when using it in one of its four different modes. You can, of course, use it as a normal laptop or a tablet but there is also the tent, where you can prop it like upside down “V” and display mode, with the screen facing up while the keyboard faces down. Weighing in at just 1.28kg, the EliteBook x360 really is a featherweight laptop that we had no trouble taking along wherever we went. While it doesn’t win any awards for being the slimmest, it just as easily slips into any bag we could throw it in. Of course with a laptop as slim as this, one must worry if HP had cut any corners with the ports in order to maintain its svelte shape. Fortunately, there are still a decent number of ports available so you won’t be trapped in dongle hell. The HP EliteBook x360 looks surprisingly stylish for a business laptop. To the right side, there is a USB 3.0 port along with a headphone jack and smart card reader for additional security. On the opposite side, there is a microSD card slot, Thunderbolt 3, HDMI output, second USB 3.0 port, Kensington lock and power jack. Right below the screen you get a full-size keyboard with backlit keys The keyboard doesn’t feel too shallow and the keys are clicky though responsive. There is also a respectably large trackpad and a fingerprint reader on the right side for added security. There is even an NFC reader hidden just beneath the trackpad. Apart from that, you get a nice stylish stylus that comes with the box. The stylus

27 Jul 2017, 3:00 PMOpenAI Newsai-labsnew item Better exploration with parameter noise

We’ve found that adding adaptive noise to the parameters of reinforcement learning algorithms frequently boosts performance. This exploration method is simple to implement and very rarely decreases performance, so it’s worth trying on any problem.

20 Jul 2017, 3:00 PMOpenAI Newsai-labsnew item Proximal Policy Optimization

We’re releasing a new class of reinforcement learning algorithms, Proximal Policy Optimization (PPO), which perform comparably or better than state-of-the-art approaches while being much simpler to implement and tune. PPO has become the default reinforcement learning algorithm at OpenAI because of its ease of use and good performance.

17 Jul 2017, 3:00 PMOpenAI Newsai-labsnew item Robust adversarial inputs

We’ve created images that reliably fool neural network classifiers when viewed from varied scales and perspectives. This challenges a claim from last week that self-driving cars would be hard to trick maliciously since they capture images from multiple scales, angles, perspectives, and the like.

05 Jul 2017, 3:00 PMOpenAI Newsai-labsnew item Hindsight Experience Replay

01 Jul 2017, 3:00 PMOpenAI Newsai-labsnew item Teacher–student curriculum learning

28 Jun 2017, 3:00 PMOpenAI Newsai-labsnew item Faster physics in Python

We’re open-sourcing a high-performance Python library for robotic simulation using the MuJoCo engine, developed over our past year of robotics research.

13 Jun 2017, 3:00 PMOpenAI Newsai-labsnew item Learning from human preferences

One step towards building safe AI systems is to remove the need for humans to write goal functions, since using a simple proxy for a complex goal, or getting the complex goal a bit wrong, can lead to undesirable and even dangerous behavior. In collaboration with DeepMind’s safety team, we’ve developed an algorithm which can infer what humans want by being told which of two proposed behaviors is better.

08 Jun 2017, 3:00 PMOpenAI Newsai-labsnew item Learning to cooperate, compete, and communicate

Multiagent environments where agents compete for resources are stepping stones on the path to AGI. Multiagent environments have two useful properties: first, there is a natural curriculum—the difficulty of the environment is determined by the skill of your competitors (and if you’re competing against clones of yourself, the environment exactly matches your skill level). Second, a multiagent environment has no stable equilibrium: no matter how smart an agent is, there’s always pressure to get smarter. These environments have a very different feel from traditional environments, and it’ll take a lot more research before we become good at them.

05 Jun 2017, 3:00 PMOpenAI Newsai-labsnew item UCB exploration via Q-ensembles

24 May 2017, 3:00 PMOpenAI Newsai-labsnew item OpenAI Baselines: DQN

We’re open-sourcing OpenAI Baselines, our internal effort to reproduce reinforcement learning algorithms with performance on par with published results. We’ll release the algorithms over upcoming months; today’s release includes DQN and three of its variants.

16 May 2017, 3:00 PMOpenAI Newsai-labsnew item Robots that learn

We’ve created a robotics system, trained entirely in simulation and deployed on a physical robot, which can learn a new task after seeing it done once.

15 May 2017, 3:00 PMOpenAI Newsai-labsnew item Roboschool

We are releasing Roboschool: open-source software for robot simulation, integrated with OpenAI Gym.

21 Apr 2017, 3:00 PMOpenAI Newsai-labsnew item Equivalence between policy gradients and soft Q-learning

10 Apr 2017, 3:00 PMOpenAI Newsai-labsnew item Stochastic Neural Networks for hierarchical reinforcement learning

06 Apr 2017, 3:00 PMOpenAI Newsai-labsnew item Unsupervised sentiment neuron

We’ve developed an unsupervised system which learns an excellent representation of sentiment, despite being trained only to predict the next character in the text of Amazon reviews.

01 Apr 2017, 3:00 PMOpenAI Newsai-labsnew item Spam detection in the physical world

We’ve created the world’s first Spam-detecting AI trained entirely in simulation and deployed on a physical robot.

24 Mar 2017, 3:00 PMOpenAI Newsai-labsnew item Evolution strategies as a scalable alternative to reinforcement learning

We’ve discovered that evolution strategies (ES), an optimization technique that’s been known for decades, rivals the performance of standard reinforcement learning (RL) techniques on modern RL benchmarks (e.g. Atari/MuJoCo), while overcoming many of RL’s inconveniences.

21 Mar 2017, 3:00 PMOpenAI Newsai-labsnew item One-shot imitation learning

Top