Abstract: "We propose a practical method for L0 norm regularization for neural networks: pruning the network during training by encouraging weights to become exactly zero. Such regularization is interesting since (1) it can greatly speed up training and inference, and (2) it can improve generalization. [...]"
Haha, thanks for dredging this up, I made that :) It seems quite related to the OP article. It's a library for making grayscale games on the TI83 graphical calculator, which has a monochrome display. The main challenge was optimizing the interrupt routine, the z80 assembly code that performed the flickering/dithering that achieves the grayscale effect, to fit within the tiny amount of clock cycles available on the Zilog Z80 6 MHz processor. Even after optimization, it took up ~50%-75% of all available cycles. Some people managed to make some pretty fun grayscale games with it (e.g. https://www.ticalc.org/archives/files/fileinfo/331/33153.htm...). This was obviously in the pre-smartphone era, so the ti83 was quite popular for playing games in class, and hand written assembly code was the only way to make it fast.
I knew immediately that the link was to Desolate before opening it :) It blew me away the first time I played it, as I was used to my shitty turn-based TI-BASIC games.
Holy crap! This takes me back. I used this (or something based on it) on my TI84 SE+ to play around with grayscale over a decade ago. It's the first thing that came to mind when I saw this article. I never got into ASM on the TI calcs but I wrote a TON of TI-Basic. I spent a ton of time on those forums and posted a number of apps/games I wrote (though I can't find them now).
Cool to see that one of the authors of the most cited paper in machine learning also got their start hacking on TI calculators. What a great learning environment that was!
Death Valley prides itself on being the hottest place on earth, based on the temperature record of 54.0 °C (air temp at 1.5m).
However, DV does not come out on top when looking at temperatures measured by satellites (EDIT: ground temperature, and less reliable.) From Wikipedia [1]: "The highest recorded temperature taken by a satellite is 66.8 °C (152.2 °F), which was measured in the Flaming Mountains of China in 2008. Other satellite measurements of ground temperature taken between 2003 and 2009, taken with the MODIS infrared spectroradiometer on the Aqua satellite, found a maximum temperature of 70.7 °C (159.3 °F), which was recorded in 2005 in the Lut Desert, Iran. The Lut Desert was also found to have the highest maximum temperature in 5 of the 7 years measured (2004, 2005, 2006, 2007, and 2009). These measurements reflect averages over a large region and so are lower than the maximum point surface temperature." According to [2], the record is 80.8 °C at Lut Desert.
You're comparing apples to oranges though. The 54.0 reading in Death Valley was measuring the temperature of the air - in particular, 1.5 meters above the ground. Satellites measure ground temperature - as in, temperature at the ground. Lots of things can affect ground temperature.
> You're comparing apples to oranges though. The 54.0 reading in Death Valley was measuring the temperature of the air - in particular, 1.5 meters above the ground. Satellites measure ground temperature - as in, temperature at the ground. Lots of things can affect ground temperature.
Exactly. I have a radio thermometer in my attic whose readings max out at 140°F. During the summer it's regularly maxed out, so the actual air temp is hotter still (and the actual surface temp of the roof must be even hotter than that). Outside air temp is nowhere near any of those temps.
You're right, it doesn't need to be an average. Meteorologists are interested in outdoor air temperature in the shade, away from local disturbances like buildings and the ground. That way it's not influenced by uninteresting local effects like the color of the walls, direct solar heating (affected by the properties of the thermometer itself) or somebody turning the oven on.
That 70.7 degrees from the satellite was ground temperature, not air temperature, so it can't be compared.
PSLDX does not dynamically adjust the leverage between stocks/bonds as a typical risk parity strategy would. Not that this is necessarily bad, this fund has a consistent exposure to a duration trade, buying long term bonds and paying short-term borrowing rate. over the last 40 years or so this has been a fantastic trade, as interest rates dropping both raises the price of bonds, propels higher equity values, and lowers the cost of leverage.
the downside to this particular fund is the extreme turnover in the fixed income component (only suitable for tax-free accounts) and the interest rate risk; the fund could underperform SPY in a world with increasing interest rates (which is where many traders believe we are now)
It does have an impressive record. 12 years is a good start, but not a long enough track record to prove much. I've been investing for nearly 40 years, and have had many with 12 good years go sour.
https://sf.gov/stay-home-except-essential-needs
"Are nannies permitted to be used? What about other in-home child care?
In some instances. If the nanny provides medical/health care for the child, then yes. If the nanny lives with you, then yes. Otherwise, nannies not necessary for medical care are not permitted. Also, parents required to work away from home for essential services may need in-home child care, which is permitted."
f. For the purposes of this Order, “Essential Businesses” means:
xviii. Home-based care for seniors, adults, or children;
h. For the purposes of this Order, “Essential Travel” includes travel for any of the following purposes.
ii. Travel to care for elderly, minors, dependents, persons with disabilities, or other vulnerable persons.
The Santa Clara(?) one that I read had some other specifics about daycare. IIRC the kids have to be in groups of 12 or fewer that aren't mixed around from day to day, and anyone in contact with one group can't be in contact with any other groups.
I just checked in on this and noticed that it seemed like the FAQ language changed as well. I can't find the old wording anymore, and the new wording has done a 180
In SF it's actually a bit more nuanced, and in most situations nannies are not allowed:
https://sf.gov/stay-home-except-essential-needs
"Are nannies permitted to be used? What about other in-home child care?
In some instances. If the nanny provides medical/health care for the child, then yes. If the nanny lives with you, then yes. Otherwise, nannies not necessary for medical care are not permitted. Also, parents required to work away from home for essential services may need in-home child care, which is permitted."
That's not the actual text of the law. The actual text says:
f. For the purposes of this Order, “Essential Businesses” means: xviii. Home-based care for seniors, adults, or children;
h. For the purposes of this Order, “Essential Travel” includes travel for any of the following purposes. ii. Travel to care for elderly, minors, dependents, persons with disabilities, or other vulnerable persons.
It says nothing about limitations to this based on need for medical care only.
Each county issued the same order. The page you cite is explanatory, but not part of an official order. Any of it that is enforceable in SF is also enforceable in the other counties.
Held individually the negative-yield bonds don't make much sense. However, they can actually improve the risk-adjusted returns of a portfolio that also holds stocks. This is because long-term bonds have, in the past decade, been negatively correlated with stocks [1,2].
https://arxiv.org/abs/1712.01312
Abstract: "We propose a practical method for L0 norm regularization for neural networks: pruning the network during training by encouraging weights to become exactly zero. Such regularization is interesting since (1) it can greatly speed up training and inference, and (2) it can improve generalization. [...]"