alexbeloi

Fun fact, they can sometimes narrow down crime scene DNA to just a single person by having enough partial matches from their (potentially distant) relatives. I can't remember which DNA database was used, but some cases were solved this way, IIRC it introduced a bunch of legal questions about if you can search a database in that way.

I think this was the article that talked about this (apologies for the paywall): https://www.nytimes.com/2021/12/27/magazine/dna-test-crime-i...

Seems like a steep price but I can see this becoming a marketable skill at some point. It feels similar to knowing how to google well, we've all internalized some google/search concepts like 'unique words' -> 'narrow results' and 'full sentences' -> 'phrase matching'. Probably there will be nuances to good gen art prompt writing.

They aren't mutually exclusive, but their incentives are opposed, usually the incentives bringing in the money win.

> What's the best way to prepare for DP in interviews?

Do 100 of these problems: https://leetcode.com/tag/dynamic-programming/

It doesn't punish anyone. If you want solutions, then the book is not for you, that's all. The author is not obligated to accommodate every audience, or even a majority audience.

I was oversimplifying, but I stand by my words.

It does optimize for profit, just with extra steps. For most FB ads products (that you see in feed), advertisers pay based on conversions (views, clicks, likes, joins, purchases, etc.). So revenue is directly tied to conversions. Then there are extra steps weighing in revenue != profit, advertiser retention, repetitiveness, long term user value, etc.

Ads optimizes for profit, all other content is broadly optimized for meaningful social interaction and against problematic content.

https://www.facebook.com/business/news/news-feed-fyi-bringin...

https://about.fb.com/news/2019/04/remove-reduce-inform-new-s...

Machine Learning is a catch-all term for optimizing statistical models with data.

Simplest example that you are very likely already familiar with is that of a 'best fit line' to some xy scatter plot. This starts by making an assumption (model choice) that the relationship between `x` and `y` is linear, e.g. `y=mx + b`, then you can use data (xy points) to figure out the most likely values for `m` and `b`. You can then make predictions for new `x_new` values by plugging them into your known line to get `y_new`.

Machine learning often manifests in a two step process: first feature extraction, and then fitting features to a desired output. Deep learning combines these as an end-to-end process to eliminate 'human in the loop' problems that occur from feature extraction.

Example: you want to predict who should win a chess game in a given board state

Feature extraction (what information you think matters): what pieces does white have, what pieces does black have, is white in check, is black in check, how many valid squares can white king move to, how many valid squares can black king move, etc...

* Fitting: make an assumption about the relationship between features and outcome (model choice), fit model using data (features, outcome)

The Deepblue 2 model that played Kasparov used around 8000 features (not sure if this is the feature vector size or # of features). As you can imagine, feature extraction is highly dependent on expert knowledge of the problem and will often fail to cover unknown situations/cases.

Deep learning models aim is to avoid limitations of expert knowledge by using raw data (e.g. occupancy of each square on a chess board) and extract features implicitly rather than relying on explicit human formulas. It has also opened up new possibilities for areas where expert knowledge has made little progress in the past (e.g. there is not much an expert can say about what pixel features are might indicate a dog/cat is contained in an image).

You'll likely be happy to hear that this has been (is being) addressed.

I watched the live broadcast of this announcement where they did a recap of all 10 previous matches (against TLO and Mana) and they talked about this concern. During today's announcement they presented a new model that could not see the whole map and had to use the camera movement to focus properly. The deepmind team said it took somewhat longer to train but they were able to achieve the same levels of performance according to their metrics and play-testing against previous version.

However...

They did a live match vs LiquidMana (6th match against Mana) against the latest version (with camera movement) and LiquidMana won! LiquidMana was able to repeatedly do hit-and-run immortal drop harassment in AlphaStar's base, forcing it to bring troops back to defend its base, causing it to fall behind in production and supply over time and ultimately lose a major battle.

Movies are filmed at 27fps so the reasoning is humans have high confidence that they aren't missing any significant information between the frames, it should possible to make a 'mental model' of a road scene at the same fps to human skill level.

In the future we'll likely have super-human spatial and temporal resolution, right now more improvements have been gained from highest possible spatial resolution with minimal plausible temporal resolution.

If by "do that" you mean mimic what a real driver would do for a specific set of sensor inputs, that is precisely ML tries to do.

To understand what the difficulty is, it's important to consider that the size of the sensor input is very large. Don't think of it like twenty range finders around the car, rather a 360 degree medium resolution color + depth image (about 0.5 million data points coming at 30 fps).

It's difficult because you will never encounter the same set of sensor inputs twice, so you can't treat it like a search space problem. Once you've accepted that, you're in AI/ML territory where you might try to reason about what the closest set of known sensor inputs and action would be (classical AI, expert system), but that is impractically difficult with as 0.5 million dimensional search space, or train an ML model to 'reason' about the sensor space to make a decision about the appropriate action.

Approaches using a small number of sensors can do automatic breaking and smarter cruise control, but haven't been seen to be successful about navigating and making strategic decisions. The current belief is that more can be done by using denser sensors and more data and seems to be the case. There are people working on reducing the sensor density requirement, but the main focus right now is building a successful and safe self driving car, regardless of sensor and compute costs.

The issues I've heard from a few people in hiring is that there is a surplus of junior data scientists from these camps and a shortage of senior data scientists to manage them. Problems not dissimilar to tech hiring in general, but companies need a lot more SWEs than data scientists.

It seems to me that Netflix used to have better recommendations. It's unclear why/how/when it went bad, my suspicion is that they changed their metric and it's having bad side effects, or their old models didn't scale as their user base grew quickly.

I've built deep learning recommendation systems in production for clients with millions of users and it's sometimes surprisingly difficult to beat the "most trending <products>" baseline if all you care about is views/purchases. It will in the short term to meet business goals, but it hurts the user experience over time and will inevitably increase churn.

That is actually one of the biggest contributing factors to the replication crisis in science, lots of scientists have been making this error for decades. Very not obvious.

The valedictorian is an 'executor', and likely very good at that crushing tasks to get the job done. The self-starter is creative and malleable and likely very good at complex problems with lots of unknowns.

Those people aren't interchangeable and it's preferable to have both to run smoothly. It sounds like you're speaking from a position of upper management, where you don't have the time to be managing an executor type person and you'd prefer more independent people directly under you.

IQ scores are also increasing[0].

Are we actually smarter than our grandparents? Or has the education system trained us to be better test takers (one of the suggested contributing factors).

[0] https://en.wikipedia.org/wiki/Flynn_effect

The relevant metrics would be daily/monthly active useres (DAU/MAU) and churn rate. But Blizzard doesn't report those number's unless they're good (e.g. Hearhstone/Overwatch hitting new record MAU last year), and hasn't ever released any such numbers for Diablo 3.

This article[0] claims Blizzards overall MAU was close to flat YoY-Q4 2016-2017, knowing that Overwatch and Hearthstone are hitting records high MAU, while overall MAU is flat means that the other games (D3, SC2, HotS) are losing players.

It's a success in the way Matrix Revolutions was a success, massively profitable[1] yet a disappointment to fans (see diablo 3 fan ratings[2]).

[0] https://venturebeat.com/2018/02/08/blizzards-monthly-active-...

[1] https://www.the-numbers.com/movie/Matrix-Revolutions-The#tab...

[2] http://www.metacritic.com/game/pc/diablo-iii

A great blog post here (https://wp.josh.com/2017/10/23/adventures-in-autorouting/) about some different auto-routing software.

Not eli5, but a comparison to the Riemann Hypothesis (RH).

RH says the Riemann-zeta function has no zeros along the line (1/2) + iy in the complex plane.

The Lindelof hypothesis says that the number of zeros between (1/2) + iy and (1/2) + i(y+1) is much smaller (little-o) than log(y) as y grows.

So it can be thought of as a weaker version of RH, but still very very difficult. The fact that Lindelof has been an open problem for over a hundred years (and is an non-trivial weakening of RH) speaks to how difficult RH is as well.

Like RH, Lindelof implies things about primes, and also (like RH) has lots of implications about lots of interesting prime-like (irreducible) objects in different spaces.

> Not completely failing to find files and apps based on how I typed them (“A” brings up “abc” but typing “ab” makes “abc” go away!?).

It's the 'basic' stuff like this and bizarre feeling UI pauses at blank windows that make it unbearable to work with. It's the opposite of snappy and constantly interrupts a productive workflow making me wonder 'why is it doing this' rather than thinking about my work. It's like having an essentially perfect phone that inexorably buzzes every 1-5 minutes (at random), you would throw it against the wall in less than a day.

Very cool, I'm a fan of his stuff and hadn't seen that one yet. Thanks

These seems like a interesting model for analyzing which population structures are most conducive to the spread of memes or ideas. The obvious population structures are 'highly connected' but this suggests that you can have loosly structured populations as long as 'have certain features (sic), such as hubs and self-loops.'

Yes they do/have research. Facebook had 12 papers at the last computer vision conference I attended (CVPR '17) and won the awards for best paper and best student paper.

My naive take on this is that it makes biologically sense to keep net expected electrical impulses approximately the same before and after strengthening. At the end of the day the brain has energy constraints.

This can-be/is done functionally in ANNs but achieves a different end (avoids over-fitting) but doesn't reducing energy(compute) expenditure in dense ANNs since activation and non-activation is computed in expectation and take the same number of cycles in dense networks.

I'd love to see more work on massive sparse networks, where you actually get compute efficiency if you can reduce number of activation without reducing hurting your optimization target.

It has little to do with culture and everything to do with current economic stage of development, Japan was known for making faulty mass produced garbage in the 50's and 60's, they used that economic model to bootstrap the manufacturing industry and to make investments in education, science, tech and manufacturing. From 1950 to 1970 Japan went from 40% agricultural to 17%. [0]

Similar story taking place in China, just 30 years later (following the cultural revolution). From 1980 to 2010 China went from 33% agricultural to <10%.[1] Now you're seeing deep investments in education, software, high tech manufacturing, science. In 20 years, it will be normal to see Chinese brands leading the market in some areas the way Toyota, Honda, Sony and Panasonic did in the 90s and 2000s (and still now in some areas).

>Name one thing we have adopted from China in terms of manufacturing principles/philosophy.

I don't know who we are in this conversation, but from my point of view, there seems to be some kind of hegemonic-cultural war going on in the JAPAC region. I have no stake in the fight, so arguments stemming from the point of view of cultural superiority will fall on deaf ears.

[0] https://en.wikipedia.org/wiki/Economic_history_of_Japan#Fact...

[1] https://en.wikipedia.org/wiki/History_of_agriculture_in_the_...

edit: A side note is that Japan in the 80s was exporting their highest quality goods to the US while leaving lower-mid quality goods for the domestic market, this is also contributed to Japan's good brand growth internationally.

Competition is welcome, deep stack knowledge from driver to model building is hard to come by (it's more than just matrix multiplies).

To compete with CUDA you need high performance drivers and people that can write them, and also be able to anticipate and keep up with the future needs of ML as well as hardware development.

>Having worked in China in the semiconductor industry, it really saddens me what the Chinese are trying to do.

I too have worked in China, and there is nothing sad about fighting and investing heavily to catch your competition. If the competition isn't careful, they'll get blindsided just like auto companies were by similar looking Japanese manufacturing investments in the 80s. Japan used to have a reputation for really low-quality manufacturing.

“While we regret that Google did not want to continue a long and fruitful tradition of collaboration between the military and technology companies, we are even more disappointed that Google apparently is more willing to support the Chinese Communist Party than the U.S. military,” they[0] wrote.

[0] The letter was signed by Republican Senators Tom Cotton and Marco Rubio, Republican Representatives Michael Conaway and Liz Cheney, and Democratic Representative Dutch Ruppersberger.

Where do the creators you follow host their videos?