Profile cover photo
Profile photo
Toby Bonang
52 followers
52 followers
About
Toby's posts

Post has attachment
Photo

Post has shared content
Visited the ITC Roxy Mas mobile phone mall in Jakarta today — mind blowing experience with huge selection, new brands I've never heard of and very aggressive prices.  In many ways, more impressive than what I've seen in China.
PhotoPhotoVideo
PhotoPhoto
Jakarta Phone Shopping
21 Photos - View album

Post has shared content
Large-scale Video Classification with Convolutional Neural Networks

With the ubiquity of videos on the internet, the development of algorithms that can analyze, summarize, and classify their content is an active field of research. While Convolutional Neural Networks (CNNs, http://goo.gl/PqYGzp) are an effective class of models for understanding image content, their development and use for video classification has been limited by the lack of video datasets that match the scale and variety of existing image datasets.

With Large-scale Video Classification with Convolutional Neural Networks (http://goo.gl/6sOs3C), a paper to be presented at the 2014 Computer Vision and Pattern Recognition conference (http://goo.gl/vY2v9k), researchers from Google and Stanford University collaborate to study the performance of CNNs for large-scale video classification using Sports-1M, a new dataset consisting of ~1 million YouTube videos belonging to a taxonomy of 487 classes of sports. 

Utilizing the information present in single, static frames as well as the complex temporal evolution present in video, the CNN learns features on full frame, low-resolution context streams along with centrally cropped high-resolution fovea streams. In doing so, the research shows that CNN architectures are capable of learning powerful features from video data, even if the provided description does not match the content, or there is significant variation on the frame level.

To support future work in this area, the Sports-1M dataset, consisting of 1,133,158 video URLs which have been annotated automatically with labels, has been made available to the research community at http://goo.gl/8imYDN. To see an example of per-frame classification results overlaid on top of a video, watch the video below.

Post has shared content
Level 3 writes: "Not only are packets being dropped, but all those not being dropped are also subject to delay. ... They are deliberately harming the service they deliver to their paying customers ... Five of those congested peers are in the United States and one is in Europe. There are none in any other part of the world. All six are large Broadband consumer networks with a dominant or exclusive market share in their local market. In countries or markets where consumers have multiple Broadband choices (like the UK) there are no congested peers ... Shouldn't a broadband consumer network with near monopoly control over their customers be expected, if not obligated, to deliver a better experience than this?"

Post has attachment
The outro to this song really reminds me of Hell's Kitchen's, with lyrics.. Both great, climactic outros.

Post has shared content
Heartbleed in pictures.

This is the most accessible explanation yet, from xkcd.

http://m.xkcd.com/1354/

#Heartbleed #xkcd #security
Photo

Post has attachment
Looks like S5 will be able to interface with Mifare chips.
That leaves LG with flagships still using Broadcomm.

Post has shared content
Introducing the latest addition to the Xiaomi Facebook family: Xiaomi Singapore. Like "Xiaomi Singapore" and we'll keep you posted.

https://www.facebook.com/pages/Xiaomi-Singapore/535405486500557
Photo

Post has shared content

Post has shared content
Fascinating speculation (presumably accurate) about Japanese web design trends.
Wait while more posts are being loaded