HNHacker News
TopNewBestAskShowJobs

heyitsguay

1,081 karma · joined January 5, 2017

submissionscomments
heyitsguay··on Adversarial Examples Are Not Bugs, They Are Features
Very agreed with all this. I've been learning the same lessons working on more robust computer vision for biomedical imaging. I bet unsupervised predictive pretraining could be adapted to (static) 3d image volumes. The z axis replaces the t axis, and you predict the next 2d slice from previous ones. Hmm...

As an aside - from the paper it looks like you worked at Brain Corp a few years back. Any thoughts on them and what they're doing these days? I'll be looking for a job again soon and i see a lot of ads for them.

heyitsguay··on Parapsychology in the PRC: 1979 [pdf]
To me that trend sounds of a kind with the slow influx of anti-vax, conspiracy theory, etc. mindsets in Internet communities; facts don't always win and the bringing together of people invested in some untrue belief provides validation and normalizes the existence of such communities.

That's not to say psi investigators deserve personal attack for their efforts, it's a harmless belief. But i think it's also ok to dismiss unsupported ideas without engaging with them. As another commenter pointed out - if there is a real phenomenon there, somebody will use it to do something.

heyitsguay··on IPFS Project Roadmap
Well and besides, the actual content on the site seems to skew alt-right pretty heavily, whether the racial slurs and swastikas are ironic or not. Taken altogether, it paints a pretty clear picture of a voat-style forum that will alienate most other people if it stays that way.
heyitsguay··on IPFS Project Roadmap
The sexual misinformation is on the front page. The racial slurs and swastikas are in the chat. Great look for a site.

I wonder what's changed for early adopters of new tech compared with the original internet in the early 90's? Early adoption was still demographically skewed towards certain groups, but i don't recall this brand of right-wing thought being so prominent.

heyitsguay··on Banned Drugs in the Meat Supply?
That has nothing to do with a blockchain though - the data structure you use has no ties to the physical entities entering data into it. If there's a regulatory body enforcing compliance, why wouldn't that regulatory body just maintain a database to track the supply chain? More energy efficient, and it aligns the data with its oversight. If there's no regulatory body, then you're back to the original problem - nothing about a blockchain prevents dirty Farm A from entering Farm B into the ledger.
heyitsguay··on Banned Drugs in the Meat Supply?
"This hamburger has a ton of crack cocaine in it"

"Oh no! To the blockchain to find out who did it!"

"Nobody recorded adding crack at any point in the supply chain"

"But blockchain is immutable and cryptographically secure!"

heyitsguay··on Sarah Hörst Creates Exoplanet Atmospheres in Her Lab
Oh yeah, absolutely! No question that it's still an issue: the vast majority of my peers had great experiences with their PI's, but I'd say 10-20% were less happy at times. Even within my department, we all knew of at least a few professors who should be avoided as advisers because of how they treated their students. So if that's anything close to a representative sample, a problem for 10-20% of people is a big problem that needs to be addressed, but that still leaves 80-90% of people pretty happy.
heyitsguay··on More jails replace in-person visits with awful video chat products
For sure, it's not up to technologists alone to solve these problems, but we should be willing to be part of the larger discussion and to learn more about the social ramifications of technological choices.
heyitsguay··on Sarah Hörst Creates Exoplanet Atmospheres in Her Lab
My grad school experience was very positive wrt my PI, and the same was true for the vast majority of my peers with whom I interacted. That's not to minimize the problem of PI's who mistreat their underlings, but to point out that in internet discussions, mostly it's unhappy people who are motivated to share their experiences, and that can create a distorted perception of [insert topic] as a whole.
heyitsguay··on More jails replace in-person visits with awful video chat products
I see this attitude a lot on HN, something along the lines of "the social problems created by this new/hypothetical technology would be hard to solve, so let's not bother trying". But as the current social media regulation fracas shows, if technologists are unwilling to engage with broader society to address the problems they create, solutions will be imposed upon them by politicians who are probably less informed about which measures are or aren't effective.

We're willing to devote a lot of time and effort to solving hard technical problems, we should be at least as willing to use some of that energy to solve the associated social problems as well.

heyitsguay··on Show HN: Thoughter
If you send a hint about a difficult conversation (#reunite, #breakup), what are the odds that the recipient won't be able to guess who sent it and the gist of the secret? Is this any better than just having the conversation directly?

If you don't send a hint, it seems extremely unlikely that the other person will use the service with the exact same hashtag, as another commenter pointed out.

So, where is the utility?

heyitsguay··on Google Stadia uses Style Transfer ML to change video game art in real time
I think the more interesting applications will come when you get tools like this used by visual artists, complementing handmade design. The demo is cool but uses style transfer very bluntly, applying one style to the whole image with high intensity. I imagine the future of this is something more like neural "shaders", able to apply real-time pixel alterations to scenes or objects within scenes. Used carefully, they could increase the range of creative choices a visual artist can make.
heyitsguay··on What's the minimum number of words you'd need to define all other words? (2012)
Looks like somebody made txt and json versions of the Oxford Unabridged English Dictionary here: https://github.com/adambom/dictionary. The json version should let build up the graph structures you're talking about pretty easily.
heyitsguay··on Biological sex shapes tumour evolution across cancer types
I don't think there are many people who sincerely believe that, it's just a strawman batted about by people who are uncomfortable with more progressive understandings of the relationship between (biological) sex and (social) gender.
heyitsguay··on Logical difficulties in modern mathematics (2012)
Oh yeah agreed, and also in other applications as well, and i tried to speak to that with the end of my comment - even with continuous time/space models, you're never going to encounter the truly pathological functions that require, e.g., the careful definitions that go into defining Lp spaces. Everything physical is much better behaved.
heyitsguay··on Logical difficulties in modern mathematics (2012)
> The users of those mathematical tools (whether software implementors or people just calling some software library) usually don’t need to care about the details of the proofs

Oh for sure, and perhaps this is just a confusion of terms but i think that's what the thread parent was speaking with "applied statistics". In academia, "applied math/statistics" can mean "I'm doing theoretical math with an eye towards applications but it still requires heavy mathematical machinery", but it can also mean "I'm using mathematical tools to solve empirical problems, and I'm never going to need to worry about Lebesgue measures".

heyitsguay··on Logical difficulties in modern mathematics (2012)
Yeah, my point is exactly what you say at the end - for practical applications, those theoretical details (which are enormously challenging) generally aren't necessary, and the problems they address aren't the ones application practitioners will encounter. I definitely understand the motivation for theoretical structures within math itself, and i have no issue with e.g. the axiom of choice for people doing mathematics itself.
heyitsguay··on Logical difficulties in modern mathematics (2012)
As someone who moved from pure mathematics -> applied mathematics -> machine learning, perhaps I can offer a similar perspective to GP that explains why the measure theory stuff might not seem so useful to some people. Basically, in application you never need to worry about anything but your simplest case - that of a discrete-time, finite-valued process. All of the subtleties of measure theory, which occupy most of what mathematical measure theorists/analysts work on, are introduced when you assume you've got an infinity or a continuum somewhere in a model, and that assumption is never borne out in practice. Unless you really want to do theoretical math, you're better off learning only the aspects that relate to finite functions, i.e. almost none of what a real mathematician would consider to be "math". And that's not just out of laziness or disinterest - once you've restricted yourself to finite functions for empirical applications, you encounter a whole new realm of engineering difficulties to deal with, that theoretical mathematics doesn't touch. You've got to learn how to solve a different class of problems that aren't as purely logical, but are every bit as challenging.

To put it another way - think about all the pathologies you may encounter working with even a "nice" function space like L^2(R). You'll never deal with those pathologies in reality, because every empirical function is much better behaved - finite domain, finite range, and even if you assume a continuous domain, you can choose a model that only has a finite number of discontinuities, is Lipschitz continuous in between them, has finite total variation, etc. And that's why the hand-wavey, "intuitive" approach works so well if you're not a theorist, at least in my opinion.

heyitsguay··on Capsule Networks- the New Deep Learning Network
Oh for sure! Not everybody needs to be trying to max out the standard benchmarks with every paper. But have there been any new developments or improvements with capsule nets since the original paper? This blog post looks like it only references that first one, which is coming up on two years old, right? It'd be great to read about any followup.
heyitsguay··on Your new smart TV is so cheap because it's collecting and selling your data
On a smart-tv I was given, I manually set garbage WiFi info and that seemed to work. No internet and no scanning.
heyitsguay··on Seminal deep learning papers
"Semen" and the "semin-" root that shows up in "seminar", "seminary", and "seminal" both come from the same Latin word for a seed, but "semin-" doesn't come from "semen".
heyitsguay··on The decline in U.S. life expectancy is unlike anything we've seen in a century
I'm having difficulty parsing this. Are you claiming that leaders of the United States are deliberately working to decrease life expectancy in the country to reduce the population?
heyitsguay··on Gradient Descent Finds Global Minima of Deep Neural Networks
Convex functions aren't the only functions with a single local (and global) minimum - consider sqrt(|x|) for a simple 1d example.
heyitsguay··on Large home libraries may have a long-term impact on proficiency: study
The relevant bit from the paper, emphasis mine:

> Cross-sectional data with retrospective indicators of scholarly culture are affected by measurement error and indicators could reflect unmeasured influence of other confounding variables which may lead to over-optimistic estimates (de Vries and de Graaf, 2008). However, prior research using longitudinal data from Australia with corrections for measurement error showed that a substantial impact of home library size on adult education existed net of the effects of academic ability (IQ) of adults or their father's scholarly habits (i.e. employment in occupations where use of books is common) as well as family income or wealth (Evans et al., 2010). Although we have no means of correcting for these potential biases in ways available in longitudinal data, we argue that even somewhat optimistic estimates are of value as an initial step in proposing and assessing hypotheses for future, more stringent tests. To limit the potential for confounding adolescent and adult literacy-enhancing activities, we control for adult out-of-work activities related to literacy, numeracy and ICT skills that might correlate with scholarly culture experienced in adolescence but are indicators of adult scholarly culture.

heyitsguay··on Artificial General Intelligence Is Here, and Impala Is Its Name
Very misleading title about a very interesting development! A link to the actual paper is here: https://arxiv.org/abs/1802.01561.

In less clickbaity terms, DeepMind has developed a scalable system for distributed training of certain types of reinforcement learning agents. They use that to train an agent to solve a pretty big collection of different reinforcement learning problems.

heyitsguay··on All-optical machine learning using diffractive deep neural networks
Can't access the article, but the abstract reads like this is inference only. If that's so, this still could be useful. For instance, you could train a network somewhere else, then implement the learned function optically.
heyitsguay··on The world is losing the war against climate change
Cancer is many diseases, that all happen to be characterized by uncontrolled cell growth. A "cure for cancer" is like a "cure for fever".
heyitsguay··on DIRT Protocol raises $3M to build a Wikipedia for structured data
Average air temperature according to whom? How do you verify the values, recorded ostensibly from sensors, were ever measured? How do you verify historical data for everything pre-2018?
heyitsguay··on Why does deep and cheap learning work so well?
A year ago, a group of us at UMD took a couple weeks to go through this paper. It's got some interesting insights, but like all theory papers right now, gaps remain. The construction they present for learning Hamiltonians of low polynomial order doesn't look a ton like any common production neural network modules, and the justification for why we would only be dealing with that Hamiltonian family in practice is unconvincing to me. That said, overall it's worth a close read. Section 2' parts A and B are the best summary of the connection between probability theory and deep learning that i have come across.
heyitsguay··on Cells Talk in a Language That Looks Like Viruses
We can in fact image them in multiple ways! Check out these super high resolution images of vesicles on a cell wall - the honeycomb bits are the vesicles' clathrin cages: https://taraskalab.nhlbi.nih.gov/media/
← PreviousPage 14 of 15Next →