Efficient large-scale fleet management via multi-agent deep reinforcement learning

Efficient large-scale fleet management via multi-agent deep reinforcement learning Lin et al., KDD’18

A couple of weeks ago we looked at a survey paper covering approaches to dynamic, stochastic, vehicle routing problems (DSVRPs). At the end of the write-up I mentioned that I couldn’t help wondering about an end-to-end deep learning based approach to learning policy as an alternative to the hand-crafted algorithms. Lenz Belzner popped up on Twitter to point me at today’s paper choice, which investigates exactly that.

The particular variation of DSVRP studied here is grounded in a ride-sharing platform with real data provided by Didi Chuxing covering four weeks of vehicle locations and trajectories, and customer orders, in the city of Chengdu. With the area covered by 504 hexagonal grid cells, the centres of which are 1.2km apart, we’re looking at around 475 square kilometers. The goal is to reposition vehicles in the fleet at each time step (10 minute intervals) so as to maximise the GMV (total value of all orders) on the platform. We’re not given information on the number of drivers, passengers, and orders in the data set (nor on the actual GMV, all results are relative), but Chengdu has a Continue reading

A quick look at QUIC

Quick UDP Internet Connection (QUIC) is a network protocol initially developed and deployed by Google, and now being standardized in the Internet Engineering Task Force. In this article we’ll take a quick tour of QUIC, looking at what goals influenced its design, and what implications QUIC might have on the overall architecture of the Internet Protocol.

Replacement Strips for Screen Privacy Filter

I use a Privacy Filter on my laptop screen when traveling. I’m doing a bit of time on planes these days, and it makes a big difference. Most of my code is Open Source, but other content is proprietary. High chance of competitors being on the same plane as me, so better to make it harder for others to see.

The only problem with these screens is that if you frequently take it off like I do, the adhesive strips collect dust, and stop sticking after a while. Recently someone asked me how to get them replaced.

3M does not sell replacement strips…but they do something even better: they give them away for free. Pretty cool ah?

Just go here, fill in the details, and they’ll send you some more. How good is that?

Replacement Strips for Screen Privacy Filter

I use a Privacy Filter on my laptop screen when traveling. I’m doing a bit of time on planes these days, and it makes a big difference. Most of my code is Open Source, but other content is proprietary. High chance of competitors being on the same plane as me, so better to make it harder for others to see.

The only problem with these screens is that if you frequently take it off like I do, the adhesive strips collect dust, and stop sticking after a while. Recently someone asked me how to get them replaced.

3M does not sell replacement strips…but they do something even better: they give them away for free. Pretty cool ah?

Just go here, fill in the details, and they’ll send you some more. How good is that?

Replacement Strips for Screen Privacy Filter

I use a Privacy Filter on my laptop screen when traveling. I’m doing a bit of time on planes these days, and it makes a big difference. Most of my code is Open Source, but other content is proprietary. High chance of competitors being on the same plane as me, so better to make it harder for others to see.

The only problem with these screens is that if you frequently take it off like I do, the adhesive strips collect dust, and stop sticking after a while. Recently someone asked me how to get them replaced.

3M does not sell replacement strips…but they do something even better: they give them away for free. Pretty cool ah?

Just go here, fill in the details, and they’ll send you some more. How good is that?

Full Stack Journey 029: The Ballerina Programming Language With Anjana Fernando

Today's Full Stack Journey podcast dances with Ballerina, a cloud-native programming language introduced by WSO2. My guest is Anjana Fernando, who has been involved in Ballerina since its inception. We discuss use cases and compare Ballerina to languages such as Java and Golang.

The post Full Stack Journey 029: The Ballerina Programming Language With Anjana Fernando appeared first on Packet Pushers.

Advanced AMI Filtering with JMESPath

I recently had a need to do some “advanced” filtering of AMIs returned by the AWS CLI. I’d already mastered the use of the --filters parameter, which let me greatly reduce the number of AMIs returned by aws ec2 describe-images. In many cases, using filters alone got me what I needed. In one case, however, I needed to be even more selective in returning results, and this lead me to some (slightly more) complex JMESPath queries than I’d used before. I wanted to share them here for the benefit of my readers.

What I’d been using before was a command that looked something like this:

ec2 describe-images --owners 099720109477 \
--filters Name=name,Values="*ubuntu-xenial-16.04*" \
Name=virtualization-type,Values=hvm \
Name=root-device-type,Values=ebs \
Name=architecture,Values=x86_64 \
--query 'sort_by(Images,&CreationDate)[-1].ImageId'

The part after --query is a JMESPath query that sorts the results, returning only the ImageId attribute of the most recent result (sorted by creation date). In this particular case, this works just fine—it returns the most recent Ubuntu Xenial 16.04 LTS AMI.

Turning to Ubuntu Bionic 18.04, though, I found that the same query didn’t return the result I needed. In addition to the regular builds of 18.04, Canonical apparently also builds EKS Continue reading

Weekend Reads 030119

The tradeoff is that users’ credentials are then centrally stored and managed, typically protected by a single master password to unlock a password manager data store. With the rising popularity of password manager use it is safe to assume that adversarial activity will target the growing user base of these password managers. Table 1, below, outlines the number of individual users and business entities for each of the password managers we examine in this paper.

Imagine that you’re about to give a presentation in front of your entire organization. If you have ever been anxious in a situation where the stakes are high, you’ve probably received well-meaning advice: Calm down. Take deep breaths. Think positively, and you will nail this. —Barry Brownstein

“Photographs furnish evidence,” wrote Susan Sontag in On Photography. “A photograph passes for incontrovertible proof that a given thing happened.” Sontag went on to write of how photographs can misrepresent situations. But do they even have to show real objects? —Ben Sixsmith

Scammers tend to be skilled at finding the most vulnerable individuals and turning them into victims. Case in point: Researchers at Proofpoint have been tracking campaigns that prey on those looking for work. The payoff Continue reading

Stuff The Internet Says On Scalability For March 1st, 2019

Wake up! It's HighScalability time:

 

10 years of AWS architecture increasing simplicity or increasing complexity? (Michael Wittig)

 

Do you like this sort of Stuff? I'd greatly appreciate your support on Patreon. Know anyone who needs cloud? I wrote Explain the Cloud Like I'm 10 just for them. It has 39 mostly 5 star reviews. They'll learn a lot and love you forever.

 

  • 1.3 billion: npm package downloads per day; 20: honeybee communication signals used to coordinate thousands of workers; 71: average global life expectancy; 120K: max inflight SQS messages; 80%: shared code between iOS, Android, the web; 1 TB: microSD card; 20%: increase in value wind energy using ML; 64%: respondents cite optimizing cloud spend as the topvinitiative; 250: drones augmenting small military units; 35,880: record robots shipped to North American companies; 50K: aerial photos of the UK; 119%: increase in demand for AI talent; 18TB: MAMR hard drive; $20 million: Pinterest paid more than expected for AWS; 100,000: MySQL connections; 19%: all requests come from Bots, APIs, and search engine crawlers; 

  • Quotable Quotes:
1 … 1,340 1,341 1,342 1,343 1,344 … 3,885