All of harshhpareek's Comments + Replies

Hi Stuart, I am about to complete a PhD in Machine Learning and would be interested in collaborations like these but only October onwards.

I have written and presented papers at Machine Learning conferences, and am quite interested in contributing to concrete AI safety research. My work so far has been on issues in supervised ranking tasks, but I have read a fair bit on reinforcement learning.

I am not close to Oxford. I am current in Austin, TX and will be in the bay area October onwards.

Are there any newbies wandering these parts? Leave a comment here! I want to know if there's any interest in a weekly/fortnightly/monthly newbie thread?

By newbie, I mean that you've found LessWrong somewhat recently and are getting exposed to many ideas like those in the Sequences but you don't post because you wonder if the things you would talk about get discussed later or might have already been discussed somewhere.

LW has a reputation of being very harsh on newbies, so maybe a newbie thread where we can discuss things without annoying those critical people would give people a place to hang out.

I've been having digestive trouble recently and have started wondering if I've developed a new allergy/intolerance (Known: milk, cashewnuts, chocolate). Does anyone have a recommendation for tests to check for these?

Apparently, "Eight foods account for 90% of all food-allergic reactions: milk, eggs, peanuts, tree nuts (e.g., walnuts, almonds, cashews, pistachios, pecans), wheat, soy, fish, and shellfish." (source: http://www.foodallergy.org/file/facts-stats.pdf). However, nuts are good for you (eg. https://examine.com/faq/how-can-i-best-ensure-ca... (read more)

I've been preparing for coding interviews, and I realized that the skill had gotten "rusty" from disuse. A specific example is coding a binary search, which is a little nontrivial because you have to think carefully to avoid off-by-one errors.

When people talk about old skills they talk about them in two ways: some skills you can supposedly never forget, like riding a bike, Some others can get rusty, so you need to keep brushing them up over and over again.

Neither of these seems actually true. I think it's more like the exponential forgetting cur... (read more)

There is an opinion expressed here, that I agree with: http://smerity.com/articles/2016/tayandyou.html TL;dr: No "learning" from interactions on twitter happened. The bot was parroting old training data, because it does not really generate text. The researchers didn't apply an offensiveness filter at all.

I think this chat bot was performing badly right from the start. It would not make sense to give too much importance to the users it was chatting with, and they did not change its mind. That bit of media sensationalism is BS. Natural language ge... (read more)

I took it.

I just attended one too! I am composing a post on this, about halfway done. I'd be interested in a collaboration where we both talk about our experiences, though I would like to see what you think. My post is laden with my own interpretations. Send me a message if you want to discuss once you have your outline down

Hi, I'm an AI PhD student and I just signed up for the Udacity Deep Learning course. Lets do this!

I'm interested in setting up the dev environment. But I'm running into technical issues setting up the VM etc. I expect more such questions will come up. What is the right place to discuss these? Perhaps a channel on the slack? Or do we want something more permanent to help new contributors?

It depends on why I'm making the list.

If I'm making a todo list for a project I'm working on, Workflowy is good because its simple and supports hierarchical lists.

For longer lived stuff where I add and delete stuff like grocery/shopping lists or books to read, I use wunderlist because they have an android app, a standalone windows app and it looks pretty. Browser-based apps annoy me so I like the windows app and the android app is nice to have when I'm actually in the grocery store.

When I'm making a list because I need to be productive and not as a way to... (read more)

I tried using RSS readers, but I tended to forget to check their websites or apps. I could have trained myself to check them more often but I ended up using https://blogtrottr.com/ instead. It sends RSS feeds to your email inbox, so I can check blogs along with my email in the morning.

I haven't had any issues so far. They send you ads along with the feed to generate revenue. Having a revenue model is a solid plus in my book.

What I don't like about it: they don't have accounts so managing subscriptions is a little hard.

I haven't put sufficient effort into identifying healthy cooking oils. I currently use Crisco's Blended Oil supplemented with Omega-3. The question is if it is supplemented in the right amount, and that information is not provided.

Animal fats are low in Omega-6 but I think the Omega-3:6 ratio is a problem for meat-eaters too.

I'm a vegetarian and I looked into this stuff a while back. The Examine.com page What beneficial compounds are primarily found in animal products? is a useful reference with sources and includes the ones you wrote above. An older page with some references is this one.

I currently supplement with a multivitamin (this one -- Hair, Skin and Nails), creatine and occasionally Coenzyme Q-10 and choline, You didn't mention the last two but I have subjectively felt they increase alertness. I (hopefully) get my Omega-3/6 fatty acids from cooking oil. I had a basic p... (read more)

Are there any LessWrongers at NIPS (in Montreal) this week? Perhaps we can have a mini meetup. Send me a PM or reply if you're here. I'm here till Sunday.

The link doesn't work anymore. Can you host it somewhere else?

I did a superficial search and it seems this is the only android app for browsing less wrong. Are there any others?

My reply is an answer to "Is there a difference between what's right and what works?" I'm not trying to open the can of worms labelled "Are draconian deterrents justified?", merely trying to show through example that what works may fall in moral grey areas. To briefly answer your comments--

If the deterrent works few hands actually need to be cut off

The effectiveness of strong deterrents is questionable#Effectiveness). Considering the links mentioned above, some cases do spring up despite the deterrent, in this particular situation. ... (read more)

Consider Saudi Arabia which even today implements the Sharia policy of cutting off the hands of those who steal, especially those who steal during prayer times. I have heard anecdotal stories, that even jewelry shops in Dubai are left unlocked during prayer times. The fear of punishment is so high that no one dares steal.

Does this policy work? Yes. Is it right? Debatable. I would argue that asking people to lock their shops is a smaller cost to society than the cost of fear and of the possible loss of limbs from this procedure, and the benefit --- being ab... (read more)

The world of the manager is one of problems and opportunities. Problems are to be managed; one must understand the nature of the problem, amass resources adequate to deal with it, and "work the problem" on an ongoing basis.[...] But what if the problem can be fixed? This is not the domain of the manager.

An engineer believes most problems have solutions. The engineer isn't interested in building an organisation to cope with the problem. [...] And yet the engineer's faith in fixes often blinds him to the fact that many problems, especially those i

I meant lukeprog's karma, i.e. the poster of a comment influences how good the comment is.

(Assuming Mind=Brain, i.e. the entire mind is just the physical brain and no "soul" is involved. Also, Neurons aren't really all that small, they're quite macroscopic -- though the processes in the neurons like chemical interactions need quantum mechanics for their description)

In Newtonian Mechanics, it is sufficient to know the positions and velocities (i.e. derivaties of position) of particles to determine future states. So, the world is Markov given this informatio.

In Schrodinger's equation, you again only need to know \Psi and it's time deri... (read more)

Not necessarily a bad algorithm. This is possible if it uses your karma as a factor, which is in general not a bad idea (in this case countered by the collapsing negative scores thing)

To develop mathematics, one must always labor to substitute ideas for calculations.

-- Dirichlet

(Don't have source, but the following paper quotes it : Prolegomena to Any Future Qualitative Physics )

Hi, I've been lurking on LessWrong for quite a while now - around a year -, but saw this post and decided to comment. I hope this is useful as feedback to the admins.

I'm a 22 year old student at UT Austin. As of last Fall, I'm pursuing a PhD in Computer Science. My specialization is Machine Learning. And I'm committed to doing everything in my power to hasten the Singularity :P. I have a BTech in CS from IIT Bombay, India.

I've considered myself a rationalist for as long as I can remember. I found less wrong through Overcoming Bias and from Elizier's posts ... (read more)

The Meander (aka Menderes) is a river in Turkey. As you might expect, it winds all over the place. But it doesn't do this out of frivolity. The path it has discovered is the most economical route to the sea

-- Paul Graham, "The Age of the Essay" (http://www.paulgraham.com/essay.html)