HNHacker News
TopNewBestAskShowJobs

xgb84j

314 karma · joined March 23, 2017

submissionscomments
xgb84j··on An IT migration corrupted 1.3B customer records
It's also the risk distribution of one large vs. many small changes that heavily favors many small changes.
xgb84j··on Attorney General will ask Zuckerberg to halt plans for end-to-end encryption
The argument about children is easily invalidated by the fact, that for some reason that during the Vietnam / Afghanistan / Irak war many children were harmed for much less than everybody's privacy.
xgb84j··on How much faster is Redis at storing a blob of JSON compared to Postgres?
The article talks about _how much_ it is faster which is the interesting question.

For a small project you might not want the added complexity of another database and decide to store JSON blobs in PostgreSQL. In this case it's important to understand the performance trade-off you're making.

xgb84j··on Ask HN: Best Linux Laptop Replacing a Macbook Pro?
I had two Dell (XPS 13 and XPS 15), both had their share of driver issues. The XPS 15 also had severe hardware issues, which support could not help with (sound randomly starting and stopping to work).

Now I use a Lenovo T480s dual-booting Windows 10 Pro and Ubuntu 18.04 and everything just works.

For Linux I recommend getting a Laptop without a dedictated GPU. Regarding Linux distributions, I stopped having issues since I stick to Ubuntu LTS releases.

xgb84j··on From Mac to Linux – the setup I've grown to love
Not gp, but I currently use a Lenovo T 480s dual booting Windows 10 Pro and Ubuntu 18.04 LTS and it's the best development setup I have used so far. No driver or hardware problems whatsoever.
xgb84j··on Ask HN: Co-Founder is lacking leadership and self-management abilities
Thank you for your reply! The help you describe is definitely an idea I can get behind. Just one question: What do you mean by ECR?
xgb84j··on Ask HN: Startup coworker underperforming
I would try to find a different role for your colleague where he performs better (e.g. tech support, creating mockups, gathering user feedback). The CEO doesn't need to directly assign him to this new role but rather give him more and more tasks and responsibilities in this area.

Also you should try to negotiate more equity with the CEO.

xgb84j··on Ask HN: Fastest way to get started with freelancing?
For me one of the best sources of work was giving a talk at a local meetup. After my first talk I had more requests than I could handle. It also helps you jump-start your network building as people will come to you after your talk to speak to you.
xgb84j··on Attention Is All You Need
When you want to process a sequence of vectors while incorporating information from a second sequence, you use attention. Attention means you create a weighted sum of all vectors in the second sequence for each vector in the first sequence. You basically add a customized summary vector of the second sequence to each vector of the first.
xgb84j··on Show HN: Compare Two Samples
Author here: I created a simple tool to compare two samples inside the browser.

The tool was created with Angular 4 and uses Universal. Currently I am thinking about creating a service that allows developers to compare software metrics of different versions of their software to detect performance regressions. But I am not sure yet how to package that as a SaaS service.

I'd be grateful for any feedback about the site (especially about how to make the whole page load faster).

xgb84j··on Writing RNNs in Tensorflow
There are techniques regime switching and indicator functions that can handle exactly this type of interaction in a linear model.

https://en.wikipedia.org/wiki/Dummy_variable_(statistics) https://en.wikipedia.org/wiki/Interaction_(statistics)

xgb84j··on Writing RNNs in Tensorflow
What I mean by this statement is that there are lots of "tricks" such as interaction terms, regime switching and non-linear transformation of features to handle non-linearities in linear models (e.g. different food sold before Christmas).

But if you can give me an example of a non-linearity in sales forecasting that cannot be fit by a linear model but can by an RNN I'd honestly be really interested in that.

xgb84j··on Writing RNNs in Tensorflow
I don't have the details of your exact use case, but I cannot imagine any complex non-linearities involved in your sales process. If ARIMA models produce decent results for your use case I would try to improve the ARIMA model through additional data, rather than switching to deep learning.

If you are convinced that there are complex non-linearities that an ARIMA model cannot describe, then I would try to use RNNs to find a pattern in your ARIMA models' residuals and try to augment your ARIMA model with an RNN.

xgb84j··on Writing RNNs in Tensorflow
I would try to include additional indicator features for weekdays or weekends directly into your Arima model.

Also I believe that RNNs are useful mainly for highly non-linear problems. Non-linearities in a problem such as sales forecasting are best handled through interaction terms or by including non-linear transformations (e.g. logarithm) of existing features into your model.

← PreviousPage 3 of 3