Posted in Neural Nets and AI Stuff on May 24, 2019
This is the first time I've heard we're able to see inside the famous "black box" of cybernetic intelligence. Maybe others have peered inside and I don't know about it, but I'm very intrigued by this line in the story. (He's talking about AND-OR-Grammar-nets):
"AOGNets are also more interpretable, meaning users can see how the system reaches its conclusions." -- Tianfu Wu, an assistant professor of electrical and computer engineering at NC State and corresponding author of a paper on the work.
"We found that hierarchical and compositional grammar gave us a simple, elegant way to unify the appr...
The best evidence I've encountered yet that the way we've learned to use binary logic is not our natural way of thinking comes out of psychology, where they have developed a trivalent way to evaluate truth values out of necessity, because people keep using logic this way, naturally:
Psychological research on people's understanding of natural language connectives has traditionally used truth table tasks, in which participants evaluate the truth or falsity of a compound sentence given the truth or falsity of its components in the framework of propositional logic. One perplexing result concerned ...
Secretary of Defense Neil McElroy said: "I want an agency that makes sure no important thing remains undone because it doesn’t fit somebody's mission." And so IARPA was created. IARPA / DARPA was involved in creating Babel and Kaldi, among other things of course.
Now I'm learning Kaldi. (Just getting started). Curiousity about the following intoductory slide led me to make this post.
This is one of those things which thoroughly surprises me, because it seems so obvious, yet it has taken me more than a couple decades to learn how to see. I've written about the idea being revealed here from a number of directions, but never before have I seen it with the clarity that I now see. The implications are significant, extending directly to Gödel's famous incompleteness proofs, and the foundations of set theory.
Here is an illustration I made to capture the idea as it presented itself to my mind this morning:
Take a moment to think about what is being described here.
As far as I kn...
Posted in Everything on May 07, 2019
Linden Gledhill, a pharmaceutical biochemist, builds custom gear that can record the beautiful, weird, and sometimes bizarre intersection of science, art, and nature. His latest photographs of cymatics, or standing waves in water, are breathtaking.
I've written about a certain little-known role of division in mathematics before, so when a friend recently posted an article I enjoyed (having posted a link to it here before), I was pleasantly surprised while re-reading to see that Cohl Furey is also intrigued by division... in a way that sorta links to my own observation. Take a look at this highlighted snippet, from this article:
Then if you're brave, take a look at my recent adventure with division in a blog post I made a month ago. It's pretty far out there. Have fun.
[Edit, a little later, after more review of Clifford Algebras, I'm fi...
Posted in Developing Software on Apr 23, 2019
If you're new to Asterisk, you can spend many hours and even days figuring out how things work before you get something working. And then as soon as you have it working, you'll get slammed by hackers trying to break in. I recently spent those hours and days getting asterisk working. So I have written a quick guide, this one for a connection to Nexmo, where you can sign up for a free account and use it as a SIP trunk to get going.
A FEW ASSUMPTIONS
Posted in Developing Software on Apr 15, 2019
The information on installing and configuring Asterisk, fail2ban, and VoipBL is all over the map. It is hilariously not easy to find what actually works. Way more confusing typos and important pieces left out on numerous sites, like there is some sort of conspiracy to make it difficult to install this trio. After trying to install Asterisk from source a number of times and failing for about a dozen different reasons, I finally gave up and used the simple "apt install asterisk" which does not install the latest version of Asterisk, but DOES install a stable, working version -- which ended up be...
Posted in Developing Software on Apr 12, 2019
After using "apt install asterisk" on Ubuntu 18.04.2, I noticed when running "service asterisk status" that there was a problem with the radius radcli configuration. It looked like this:
radcli: rc_read_config: rc_read_config: can't open /etc/radiusclient-ng/radiusclient.conf: No such file or directory
The full stack trace looks like this:
root@:/etc/asterisk# service asterisk status ● asterisk.service - Asterisk PBX Loaded: loaded (/lib/systemd/system/asterisk.service; enabled; vendor preset: enabled) Active: active (running) since Fri 2019-04-12 11:03:11 CDT; 32s ago Docs: man:asterisk(8) Ma...
This peculiar thought experiment captures the first time I've been able to completely flip the idea of counting, with a rather original approach. Other attempts have been more high-level, with fewer details, and couldn't be pursued very far. But this time I got to the bottom. Please be patient with this idea. It gets messy at points, but I think the underlying restructuring can be done, even if only as a novelty.
Note, I do not think my proposal below is a sensible one. It is composed of pieces which don't yet fit together well -- and one of them is completely absurd -- but I'm leaving it in ...