Friday Fun (Thanksgiving 24-Nov-2017 Edition)

  • Can an AI be taught to explain itself? Cliff Kuang, New York Times Magazine
    This is a good account of some of the problems we face with machine learning today. There is a clear disconnect between the results you get with good applications of ML, and understanding why they work the way they do. I am not convinced, however, that just adding a second network on the side to explain the first really will solve the problem — it begs the question of how we will understand what that network is doing.
  • Come On Eileen, Dexy’s Midnight Runners. It’s worth finding different versions of this song and listening, because there are some fun intros and exits you don’t hear on the usual radio mix! See the wikipedia page for a nice discussion.

Friday Fun (17-Nov-2017 edition)

Friday Fun (10-Nov-2017 edition)

  • Sixty Years of Software Development Life Cycle Models, Kneuper, Ralf. IEEE Annals of the History of Computing. The Hegelian account of software development life cycles is apparent to anyone who’s been around for more than a decade, or even worked in different sectors of the industry. In my mind, what Kneuper brings to the discussion in this case is not a simple account of the thesis, antithesis, and synthesis of software development life cycles, but interesting facts about their early development. Prototypes played a role much earlier in lifecycle planning than I think many have been aware of, as was an iterative approach with feedback loops in general.
  • The Worst Day Since Yesterday, Flogging Molly. It’s been that kind of a week around here. I highly recommend you go out, get a Guiness, and crank up Flogging Molly as loud as your speaker will allow. You can’t go wrong with that on a Friday evening.

Friday Fun (03-Nov-2017 edition)

  • Idea of Order at Kyson Point, Brian Eno. Brian Eno needs no introduction; this is a nice short recent work he put out this year.
  • Deep Reinforcement Learning: Pong from Pixels. As promised, here’s a bit of a flashback on reinforcement learning, a neat older result on using reinforcement learning to train a network to play Atari video games. It’s important to recognize in this work, too, just like with the AlphaGo Zero work, that the resulting network does not understand what it’s doing. It can’t explain the rules, doesn’t have any abstractions. It’s just very, very, very good at pattern recognition.

Friday Fun (27-Oct-2017 edition)

  • More Than, Au Revoir Simone. Dreamy synthpop at its very best.
  • Mastering the game of Go without human knowledge, David Silver, et al; good Nature summary coverage as Self-Taught AI is best yet at Strategy Game of Go, Elizabeth Gibney. This is a very important result, although I think it’s been a little too widely hyped by the popular press as evidence of the coming singularity. Go is an interesting problem domain, because the combinatorial explosion of movies leaves it intractable for traditional game-playing approaches. Reinforcement learning, used by the team, is essentially how humans learn to play go, albeit far, far faster than we learn to play. I am looking forward to seeing discussions in the coming months of the new strategies AlphaGo Zero teaches human players.

Friday Fun! (20-Oct-2017 Edition)

(So, yeah, last week’s promise of a post tomorrow didn’t quite pan out. Anyway, without further ado…)

  • Resonant Expanse, Max Cooper & Tom Hodge
    I love almost everything I’ve heard by Max Cooper. He takes traditional trance to a whole new level with his use of modulation on minimalist melodies and percussion. This work by he and Tom Hodge is on several of my “music to think by” playlists.
  • A Preliminary Analysis of Sleep-Like States in the Cuttlefish Sepia officinalis, Marcos G. Frank , Robert H. Waldrop, Michelle Dumoulin, Sara Aton, Jean G. Boal.
    The punch line is in the abstract: “In addition, cuttlefish transiently display a quiescent state with rapid eye movements, changes in body coloration and twitching of the arms, that is possibly analogous to REM sleep.” Cephalopods and mammals diverged some five hundred million years ago — like, twice as long ago as when dinosaurs and mammals were hanging together. If this holds true, it’s amazing. I can’t even really call it convergent evolution, because I’m not convinced we can articulate what evolutionary pressures would generate the need for REM sleep in such different ecosystems, unless it’s actually a requirement for brain function. But human and cephalopod brains are very, very different — the common ancestor was probably something like a sea worm with a brain similar to C. elegans. So there’s an awful lot of room for divergent evolution, which we see in things like the gross structures.

Anyway, thinking of cuttlefish and perhaps octopuses dreaming of Max Cooper’s and Tom Hodge’s music makes me very, very happy.

New! Friday fun!

For years I’ve been in the intermittent habit of occasionally sending coworkers interesting things on Fridays that I find on the web. It may be a paper, blog post, or newspaper story that I found particularly interesting that they might not have encountered. Often, but not always, tech related. It might be a paper on ML, a history of computing paper, a blog post on keeping organized or using agile, or something like that. More recently, I’ve added a link to a single song I’d like to share. With both of these I provide a bit of commentary — no more than a few sentences, explaining why I thought this was worth following along.

I’m going to start cross-posting the content here, because there’s never anything proprietary about it, and it occurs to me that it might be of interest to a broader audience.

One remark is in order. I make no apologies that some of what I reference is behind one paywall or another, and no, I won’t send copies of what I link if you can’t access it. I am a member of the IEEE, ACM, and ARRL in part because I value the editorial work they do in their journals and to have access to their digital libraries; the fee I pay for membership in part enables them to do the editorial work that they do. The same goes for the few online news sources I pay fees to access. In all cases, there are ways for motivated non-members to access the works these organizations provide or curate — typically there’s free access for a number of works a month, or the opportunity to buy a reprint of a particular paper at a nominal cost. If what you say really interests you, it’s worth your effort to support the sources that curate that material for you.

Look forward to the first post tomorrow!

SLV Winlink Packet Survey

With help from KE6AFE and WB6RJH, I recently installed a Winlink RMS Packet gateway at the home station in the hopes that having easy access to one would enable me to regularly test my portable packet station — in the past I would assemble one for a specific ARES or public service deployment with bits and pieces, test it, use it for the deployment, and then pieces would be cannibalized for other projects, or software would be reinstalled. Although the topology at the house is not ideal for providing RF coverage, I thought that it would at least be a good exercise, and once set up, I could see if it added anything to the local network.

(I’ll write more about the station configuration in a later post; for this discussion it’s enough to know that it’s connected to a Kenwood TH-D700A radio running 50W into a j-pole on top of the house, but unfortunately below the lip of the canyon we live in.)

San Lorenzo Valley is served by two digipeaters, WR6AOK-3 above the middle of the valley on 145.690, and W6JWS-3 to the northeast on 145.630. In addition, WB6RJH operates WB6RJH-10 on 145.690 above the south half of the valley, with coverage extending into parts of Santa Cruz as well as most of Bonny Doon. Although not in San Lorenzo Valley, W6SCF-10 operates on 145.630, and is accessible from W6JWS-3.

Conferring with KE6AFE, we thought the best thing for me to do was put KF6GPE-10 at my home location running on 145.630, so that’s what I did. Over the month of August, the three of us performed a number of tests at locations known to be of interest during ARES activation, with a bias on Boulder Creek locations because that’s where I live and was most curious about possible simplex paths between Boulder Creek and KF6GPE-10.

I used a Kenwood TH-D72A HT running at 5W to the low-profile magmount antenna on my mobile station for the experiments.

Results were quite good, with redundant relays available at each location. I still need to exercise the WR6AOK digipeater more in both Ben Lomond and Boulder Creek; KE6AFE and WB6RJH provided me with a KA-Node script for relaying through WR6AOK that is more reliable than using plain AX.25 digipeating, but I’ve only tested that at one location. Knowing the regions that remain to be tested it does appear that each served agency location is likely to have access to at least two packet relays through at least two different paths on two different frequencies. Unfortunately, there aren’t as many simplex paths as I would like — as I feared, KF6GPE-10 is essentially unavailable by simplex from anywhere likely to be interesting (other than my living room and Redwood Elementary), and WB6RJH is far enough to the south that simplex at low power is not an option for locations very far north of Felton.

Without further ado, here are the results.
Continue reading SLV Winlink Packet Survey

Swiss Micros DM41 Quick Reference

Now that I’ve been using the excellent Swiss Micros DM-41 calculator a while, there are a few things I still tend to forget that are either different than the HP-41CX, or things that are the same that I just never did often enough. The HP-41CX Quick Reference Guide (if you can find one) is really handy, and I’m lucky enough to have an original printed copy in my notebook’s pocket. But there are some specifics to the DM-41, notably the power-on key assignments, that are helpful to have handy.

And thus is born my DM-41 Quick Reference Guide, available under a Creative Commons Share Alike Attributions (CC-BY-SA) license.

DM41 Quick Reference

It’s formatted to print for trifold to business size, so you can slip it in the business card elastic holder of the DM-41 flip cases.

You can download and make your own from this PDF. I’ve done a limited print run on card stock; if you’re interested in getting one by sending me a SASE, do let me know by email or Twitter (@kf6gpe).

Edit 20-Aug-2017: Updated to indicate that the cards are available for an SASE.