Connect with us

Future-Guided Incremental Transformer for Simultaneous Translation

More Videos

Simultaneous translation is the type of machine translation, where output is generated while reading source sentences. It can be used in the live subtitle or simultaneous interpretation.

However, the current policies have low computational speed and lack guidance from future source information. Those two weaknesses are overcome by a recently suggested method called Future-Guided Incremental Transformer.

Image credit: Pxhere, CC0 Public Domain

It uses the average embedding layer to summarize the consumed source information and avoid time-consuming recalculation. The predictive ability is enhanced by embedding some future information through knowledge distillation. The results show that training speed is accelerated about 28 times compared to currently used models. Improved translation quality was also achieved on the Chinese-English and German-English simultaneous translation tasks.

Simultaneous translation (ST) starts translations synchronously while reading source sentences, and is used in many online scenarios. The previous wait-k policy is concise and achieved good results in ST. However, wait-k policy faces two weaknesses: low training speed caused by the recalculation of hidden states and lack of future source information to guide training. For the low training speed, we propose an incremental Transformer with an average embedding layer (AEL) to accelerate the speed of calculation of the hidden states during training. For future-guided training, we propose a conventional Transformer as the teacher of the incremental Transformer, and try to invisibly embed some future information in the model through knowledge distillation. We conducted experiments on Chinese-English and German-English simultaneous translation tasks and compared with the wait-k policy to evaluate the proposed method. Our method can effectively increase the training speed by about 28 times on average at different k and implicitly embed some predictive abilities in the model, achieving better translation quality than wait-k baseline.

0

Nokia 1.4, Nokia 6.3, and Nokia 7.3 May Launch in Late Q1 or Early Q2 This Year

Nokia 1.4, Nokia 6.3, and Nokia 7.3 could launch in Q1 or early Q2 of this year, a new report claims. All three of these Nokia phones have been in the news in the past with speculations around their release. While Nokia 1.4 is relatively new, Nokia 6.3 and Nokia 7.3 were originally expected to launch in the third quarter of 2020. It is also possible that whenever these two Nokia phones launch, they could also be named Nokia 6.4 or Nokia 7.4.

Starting with Nokia 1.4, the phone has made its way through multiple listings, as per a report by Nokiapoweruser. The report states that the phone may launch in February. Recently, specifications and pricing for Nokia 1.4 were tipped, suggesting a 6.51-inch HD+ LCD display, a quad core processor, 1GB + 16GB storage configuration, and dual-rear camera setup. The phone is expected to be priced under EUR 100 (roughly Rs. 8,800).

Nokia 6.3 and Nokia 7.3, on the other hand, have been in the news for quite a while now. The report by Nokiapoweruser states that they may launch late in the first quarter or early in the second quarter of 2021. These phones could also launch as Nokia 6.4 and Nokia 7.4.

Nokia 6.3 has been tipped in the past to come with the Qualcomm Snapdragon 730 SoC and a 24-megapixel shooter. Nokia 7.3 may feature a 6.5-inch full HD+ display with a hole-punch cutout and be powered by the Qualcomm Snapdragon 690 SoC. It could come with a 48-megapixel primary sensor and a 24-megapixel selfie shooter. Nokia 6.3 and Nokia 7.3 are expected to be backed by a 4,500mAh and a 5,000mAh battery, respectively.

Originally, the two phones were expected to launch at IFA 2020 in September. Then, it was reported that they may launch in November. Even now, Nokia or brand licensee HMD Global has not shared any information on the launch date for these phones.

Is Android One holding back Nokia smartphones in India? We discussed this on Orbital, our weekly technology podcast, which you can subscribe to via Apple Podcasts, Google Podcasts, or RSS, download the episode, or just hit the play button below.

0

Linux 101: How to copy files and directories from the command line

Jack Wallen continues his Linux 101 series, with an introduction on how to copy files and directories from the command line.

Are you new to Linux? If so, you’ve probably found the command line can be a bit intimidating. Don’t worry–it is for everyone at the beginning. That’s why I’m here to guide you through the process, and today I’m going to show you how to copy files and folders from the command line.

Why would you need to copy files and folders this way? You might find yourself on a GUI-less Linux server and need to make a backup of a configuration file or copy a data directory.

Trust me, at some point you’re going to need to be able to do this. Let’s find out how.

SEE: Linux: The 7 best distributions for new users (free PDF) (TechRepublic)

First we’ll copy a file. Let’s say you’re about to make changes to the Samba configuration file, smb.conf and you want a backup copy just in case something goes wrong. To copy that file, use the cp command to copy the source to the destination like so:

 cp /etc/samba/smb.conf /etc/samba/smb.conf.bak

You’ve probably already encountered your first problem. Because the smb.conf file is in /etc/, you’ll need to use sudo privileges to make the copy. So the correct command is:

sudo cp /etc/samba/smb.conf /etc/samba/smb.conf.bak

In this example, smb.conf is our source and smb.conf.bak is our destination. You might want to preserve the file attributes (such as directory and file mode, ownership, and timestamps) during the copy. For that we use the -a option as in:

sudo cp -a /etc/samba/smb.conf /etc/samba/smb.conf.bak

Copying a directory is done in the same way, only you use the -R option, for recursive. Let’s say you want to make a backup of the entire /etc/samba directory and you want to copy it to your home directory. That command would be:

sudo cp -R /etc/samba ~/samba.bak

To preserve the attributes, while copying the directory, the command would be:

sudo cp -aR /etc/samba ~/samba.bak

And that’s all there is to it. You’ve just copied your first files and directories from the Linux command line. Now, go out and celebrate this victory, you’ve earned it.

Subscribe to TechRepublic’s How To Make Tech Work on YouTube for all the latest tech advice for business pros from Jack Wallen.

0

‘Junk DNA’ plays a key role in regulating circadian clocks

If you’ve ever had a bad case of jet lag, you know how a disruption to your body’s circadian rhythm makes it difficult to function. Molecular circadian “clocks” exist in cells throughout the body, governing more than just sleep and wake cycles — they are crucial to many aspects of human health. For more than a decade, researchers have been trying to figure out what makes them tick, in search of new insights into diseases like Alzheimer’s, cancer and diabetes.

Until now, that research has focused on what is known as clock genes, which encode proteins that drive oscillating cycles of gene expression affecting physiology and behavior. But research just published in the Proceedings of the National Academy of Sciences reveals the discovery of a new cog in the circadian clock — a genome-wide regulatory layer made up of small chains of non-coding nucleotides known as micro RNAS (miRNAs).

Drosophila ciacadian rhythm. Credit: Chhandama via Wikimedia Commons, CC-BY-SA-4.0

“We’ve seen how the function of these clock genes are really important in many different diseases,” said Steve Kay, PhD, Provost Professor of neurology, biomedical engineering and quantitative computational biology at the Keck School of Medicine of USC. “But what we were blind to was a whole different funky kind of genes network that also is important for circadian regulation and this is the whole crazy world of what we call non-coding microRNA.”

‘Junk DNA’ proves to be a valuable tool in circadian rhythms

Formerly thought to be “junk DNA,” miRNAs are now known to affect gene expression by preventing messenger RNA from making proteins. Past research has indicated miRNAs may have a role in the function of circadian clocks but determining which of the hundreds of miRNAs in the genome might be involved remained a problem.

Kay and his team, led by Lili Zhou, a research associate in the Keck School’s Department of Neurology, turned to the Genomics Institute of the Novartis Research Foundation (GNF) in San Diego, which has created robots capable of high throughput experiments. Working with scientists at the institute, Zhou developed a high throughput screen for a robot to test the close to 1000 miRNAs by individually transferring them into cells the team had engineered to glow on and off, based on the cell’s 24-hour circadian clock cycle.

“The collaboration with GNF made it possible for us to conduct the first cell-based, genome-wide screening approach to systematically identify which of the hundreds of miRNAs might be the ones modulating circadian rhythms,” said Zhou.

“Much to our surprise,” said Kay, “we discovered about 110 to 120 miRNAs that do this.”

With the help of Caitlyn Miller, a biochemistry undergraduate from USC Dornsife, researchers then verified the impact on circadian rhythms by inactivating certain miRNAs identified by the screen in their line of glowing cells. Knocking out the miRNAs had the opposite effect on the cells’ circadian rhythm as adding them to the cells.

Physiologic and behavioral impacts  of miRNAs

Researchers also focused on the physiologic and behavioral impacts of miRNAs. They analyzed the behavior of mice with a particular cluster of miRNAs inactivated – miR 183/96/182 – and saw that inactivating the cluster interfered with their wheel-running behavior in the dark compared with control mice. They then examined the impact of the miRNA cluster on brain, retina and lung tissue, and found that inactivating the cluster affected circadian rhythms in a different way in each tissue type – suggesting that the way the miRNAs regulate the circadian clock is tissue specific.

Understanding the impact of miRNAs on the circadian clock in individual tissue could reveal new ways of treating or preventing specific diseases.

“In the brain we’re interested in connecting the clock to diseases like Alzheimer’s, in the lung we’re interested in connecting the clock to diseases like asthma,” said Kay. “The next step I think for us to model disease states in animals and in cells and look at how these microRNAs are functioning in those disease states.”

Source: USC