Ouf ! des tips pour butter les robots tueurs
RT @Gritty20202
Tips for dealing with robotic police dog (SPOT)

this just seems like valuable information

Whoever thinks they're going to buy BTC at $28k is very optimistic.

The shape of these two curves implies upward pressure on the price since there are way more "buy" orders than "sell".

How is it that the Enterprise went the entire first season of TNG without a consistent chief engineer?

Maybe for the same reason Roddenberry was concerned that they'd never build the main engineering set if he didn't include scenes there in the pilot?

I’ve started making heavy use of β€œhide boosts” so hopefully you post some original content too.

Zooko, famously of Zooko's Triangle says "This is a great upgrade to the original Pet Names paper!! Everyone who works on decentralized systems and/or social networks should read this! ❀️"


Can't get better praise than someone famous for analyzing naming properties talking about your paper about naming things!

Why didn’t they just leave Moriarty running so he could research ways to get off the holodeck on his own?

Former Malware Distributor Kape Technologies Now Owns ExpressVPN, CyberGhost, Private Internet Access, Zenmate, and a Collection of VPN β€œReview” Websites


> In 1670, a jury at the Old Bailey declined to obey the judge’s direction to convict two Quakers, William Penn (1644–1718) and William Mead, despite having them locked up for days without food or fire or chamber-pot. The Court of Common Pleas, who heard the jury’s appeal, was forced to acknowledge that the right to trial by one’s peers, as stated in Magna Carta, entailed a right to acquit, irrespective of the judge’s view that the defendant was guilty.

It literally has been a right since the Magna Carta

They should untar them and see what's inside.

"Tar balls possibly linked to California oil spill are appearing on San Diego beaches..."


Data is more important than algorithms? 

This does nothing at all to make it easier to train the model that "GlotNet" and the vocal tract filter use as input, but it does show that with clever design of the end-to-end system one can get good results with less data. "The Unreasonable Effectiveness of Data" doesn't say you can't do this; just that if you have enough data you don't have to.

Now I need to figure out how to split up the piece that produces the acoustic features.


Show thread

Data is more important than algorithms? 

Fortunately, it appears that this is indeed doable. I found a paperΒΉ in which the authors use a WaveNet conditioned on "glottal excitation" instead of the final output waveform, then use the output to excite a vocal tract filter. The glottal excitation WaveNet and vocal tract filter both use the same speaker-dependent acoustic features as input, but together they produce better results than WaveNet with manageable amounts of data.

ΒΉ arxiv.org/pdf/1804.09593v1.pdf

Show thread

Data is more important than algorithms? 

We already do this far up in the stack: the components that translate text to phoneme sequences and generate pitch trajectory and prosody information are all speaker-independent. If there's a vocoder involved, it will be speaker-independent as well. But there are usually 1-2 intermediate stages that are speaker-dependent, and these stages require a lot of training data.


Show thread

Data is more important than algorithms? 

I'm wondering, though, if it might be possible to make up for insufficient data by using "stacks" of models where the data you do have is large enough for the particular sub-task.

In other words, for something like speech synthesis, could one separately train some set of speaker-independent models, then feed those into a speaker-dependent model to produce the final output?


Show thread

Data is more important than algorithms? 

Pre-trained models you can't reproduce are the equivalent of proprietary software. But most users of free software don't bother compiling it themselves, so I think it's fine to use a pre-trained model as long as the license is OK and everything you need to reproduce the model is at least available, even if you don't actually reproduce the model.


Show thread

Data is more important than algorithms? 

I've started reading a book on machine learning, and one of the things it points out early on is that with enough data, all ML algorithms perform roughly the same. This sucks, because it means there's only so much a lone tinkerer like me can do.

One of the ways to deal with this problem is to use pre-trained models that can be fine-tuned with manageable amounts of additional data.


Subtoot: If you're engaged in a public thread, and someone responds mentioning you ... well, that's kind of how the mechanics of Mastodon / the Fediverse work.

If the interaction is otherwise not directly hostile, you can:

Indicate the information wasn't relevant / helpful, and/or how.
Ignore the toot entirely.
My fave: just give a "thanks". It's an acknowledgement, but not an invitation. If not heartfelt, well, you know that, but you don't look like the asshole either.
Or, you know, you can go off on a tear on the schlub.

But point remains, you were talking in public, and the public responded.

If you want to have discussions not in public, you might want to reconsider where and how you're discussing things.

And if you're on the receiving end of a tear / unfriendly response, a friendly "sorry, thought it might be relevant" might also be a bit of needed social lubricant. The apology tends to soften the defence.

I've been on all three (or four) sides of this myself: respondee, responder, the public, other thread participants.

I find that the deflecting approaches seem to work best (and if the counterparty is really an asshole, well, that tends to reveal itself in short order).

Stress is contagious. Deflections help reduce it for everyone.

"Be kind, for everyone you meet is fighting a hard battle"


#MastoTips #PublicIsShared

Some people consider this to be evidence that the US is evil, but despite the fact that the US had a big hand in the creation of the UDHR, the American concept of rights is very different from the European one. In particular Americans for the most part believe only in *negative* rights, i.e. the right to be left alone. We're a lot less comfortable with rights that impose obligations on others, like a right to "free" health care or whatever.


Show older