Recent searches

No recent searches

Search options

Only available when logged in.
mastodon.world is one of the many independent Mastodon servers you can use to participate in the fediverse.
Generic Mastodon server for anyone to use.

Administered by:

Server stats:

8.2K
active users

mastodon.world: AboutStatusProfiles directoryPrivacy policy

Mastodon: AboutGet the appKeyboard shortcutsView source codev4.4.0

#PCA

0 posts0 participants0 posts today
Habr

Главное по ML/DL, часть 2: Вопрос → Краткий ответ → Разбор → Пример кода. SVD/PCA. Bias-variance. Деревья. Бустинг

У каждого наступает момент, когда нужно быстро освежить в памяти огромный пласт информации по всему ML. Причины разные - подготовка к собеседованию, начало преподавания или просто найти вдохновение. Времени мало, объема много, цели амбициозные - нужно научиться легко и быстро объяснять , но так же не лишая полноты! 💻 Обращу внимание, самый действенный способ разобраться и запомнить - это своими руками поисследовать задачу ! Это самое важное, оно происходит в секции с кодом. Поэтому попробуйте сами решить предложенную задачку и придумать свою! Будет здорово получить ваши задачи и в следующих выпусках разобрать! Мы продолжаем. Обязательно испытайте себя в предыдущей [1] части! В лес, так в лес!

habr.com/ru/articles/921190/

ХабрГлавное по ML/DL, часть 2: Вопрос → Краткий ответ → Разбор → Пример кода. SVD/PCA. Bias-variance. Деревья. БустингУ каждого наступает момент, когда нужно быстро освежить в памяти огромный пласт информации по всему ML. Причины разные - подготовка к собеседованию, начало преподавания или просто найти вдохновение....
#machinelearning#ds#python
Greg Johnson

Eliphalet Nott was Presbyterian minister and RPI (Union College) President. He says many wish they could serve Jesus like Martha (Luke 10:38-42). Yes, Jesus has ascended above beyond our physical reach. But Matthew 25 says that kindness to the pious poor is the same done to Christ.

Today, many seek extrabiblical ways to express zeal for service to “Christ himself”.

How can you open your doors to the Son of David?

#christian#pca#wokeism
*
screwlisp

#programming #engineering #statistics #PCA in #commonLisp #lisp #blog #easy #reference screwlisp.small-web.org/progra

While I am just feeding no-other-obvious-source lisp pieces of my brain to my young kitten, here is principle component analysis in common lisp using an openly available #ML package #clml from a Japanese telco changing hands 15 years ago, actively developed by a lisp lone wolf up until five years ago.

My demo is in english (#emacs #eev #eepitch) in contrast to the Japanese internals.

screwlisp.small-web.orgLisp Principal Component Analysis (PCA) (eigendecomposition)
ferricoxide

Cramming for my #GCP #PCA exam, tomorrow. Googling around to see if I can find some practice tests to work through to expose any gaps in the knowledge I'll need for the test (the Google-provided ones were "too easy"). Found this one site, but it really feels like:

* The person(s) writing the quiz may be ESL
* The person(s) writing the quiz lack attention to detail (see attached image)
* Some of the questions' answers are either straight-up wrong or not as accurate as they ought to be

I assume that at least some of them are "someone sat the test, then tried to write down what they recalled from the tests". But still…

ferricoxide

Continuing to work through #GCP training – having gotten my #ACE, I now need to get my #PCA.

At any rate, one of the training-modules was talking about data transfer times and popped up a size/rate table to show transfer-times. It instantly transported me back to when I was doing DR-related delivery-consulting.

I still remember one customer being pissed when I'd finished setting up their storage-replication and declared, "well, I'll see you in about 21 days to verify that we're actually finished and functional". They were incredulous, sputtering about "how can you possibly say 21 days". I'd reminded them that I'd noted that their replication-bandwidth was objectively too small (while they had a T3, their data set was large and they couldn't get their networking people to remove the session-limits that were constraining the replication to a fraction of that T3's capability). We'd spent a couple days benchmarking and working with their networking people to try to address the discrepancy between their theoretical bandwidth and the observed bandwidth. In fairness, they probably would have been ill-advised to dedicate the entirety of the T3's bandwidth to replications — since they presumably had other projects that needed some of its bandwidth — but the amount they were able to allocate to the storage replication was way too low. I'd given them an initial transfer-time estimate, the day prior, based on my benchmarking and the first couple hours of the replication's sync-up. My return the next day was mostly to confirm that things were moving along as expected — so that I didn't have to revise my estimate (fortunately, over the intervening 16 hours, my numbers had stayed dead-on).

Just to add some gas to the fire:

• I reminded them that they had had the option to speed up the initial sync by seeing it with a tape-based restore – which would have required them to ship tapes from NJ to AZ, import the tapes into their DR site's tape library system, then do a restore to the DR site's storage array
• I pointed out to them that, with such a low transfer-rate, if they ever had link-loss, it could take them hours to days to get back in sync. I further pointed out and that, worst case (i.e., on a long-enough link-outage), the replication-software might declare the sync "stale" and they'd have to re-initialize and re-do the transfer.

This was probably 2005 or 2006. So, it's not that they had a ton of data to transfer, it's just that cross-country private circuits weren't available with nearly as high of transfer speeds and what options were available were silly-expensive.

TrendingLive feeds
About