HN user
djoshea.com
How does Zotero compare to Paperpile these days? When I first found Paperpile, I was astonished how seamless the browser extension based citation and PDF auto-import worked and never looked back. But I imagine Zotero has come a long way as well?
Congratulations Lachy and the π team! This strikes me as a guide star for neuroscience (for me at least): understanding how the brain achieves physical intelligence. Clearly our brain learns and masters skills by distilling and transferring knowledge about how to interact with the physical world. Some of the methods your team are developing point towards algorithms and representations to search for in the brain. Exciting stuff!
The Simons Foundation has had an enormous, transformative impact on neuroscience as well. It’s widely considered among the most incisive, forward-looking sources of funding in the field, pushing for fundamental advances to solve “tomorrow’s problems.” https://www.simonsfoundation.org/collaborations/#global-brai...
I used pipes back in college to randomize our roommate ordering in picking dorm rooms. I had it take the top headline on The NY Times website, hash it, and pull out numbers from 1-6 in a deterministic way. It was the best way I could think of to do it from afar over the summer, and convince everyone it was fair without needing to run code ourselves. It was useful and fun to use!
Years ago I wrote an oscilloscope-style monitor for an ATI load cell using ImGUI in the early days. This library would have been extremely useful. Kudos!
Screenshot: https://cloud.githubusercontent.com/assets/77752/5382993/9f6...
Summary from the abstract with acronyms expanded:
“In sum, we have identified NT (neurotensin) as a neuropeptide that signals valence in the BLA (basolateral amygdala) and showed that NT is a critical neuromodulator that orchestrates positive and negative valence assignment in amygdala neurons by extending valence-specific plasticity to behaviourally relevant timescales.”
Not really. Lots of neurons in this part of the brain modulate their activity to movement in heterogeneous ways. The algorithm details vary, but at some level you’re trying to find a 2d x/y velocity signal encoded in the 200d neural signals. This decoder is a bit more sophisticated (using deep learning style approaches), but a Kalman filter was state of the art for a long time.
For the adaptation, there’s a rich literature of neuroscientists in this field studying how the participant adapts to the control characteristics of the decoder, and how the decoding algorithm can be designed to adapt after seeing more data during use. Here’s one paper if you’re interested http://www.stat.columbia.edu/~liam/research/pubs/merel-fox-c...
Does anyone know of a way to guard rclone passwords that would be stored in rclone.conf? Is it possible to leverage the keychain or this tool for this? Typically passwords for sftp logins are stored with a reversible hash (see `rclone reveal`), which isn't ideal.
Exactly. I believe the 60 million infections delta was estimated for April 6th (figure 4), so the counterfactual would be considerably worse as of today.
As their website is crawling to a halt, there’s also this story on the Harvard Crimson. https://www.thecrimson.com/article/2019/12/3/grad-union-stri...
If you're not specifically looking for Computer Modern fonts in your figures but just want vector graphic WYSIWYG exports with your pick of font (beyond what Matlab renders natively), my Matlab figure export tool works well:
They're not transcribing it, so much as selecting the keys one would need to press in order to type the text. That is, the correct key lights up, and they move the cursor and mentally "click" to select it. This serves two purposes:
- WPM is a difficult metric to "game", since if they make a mistake they have to select the backspace key, and then select the correct key again.
- It's a proof of concept demonstration that a human (e.g. in the BrainGate[1] clinical trial) could use this prosthetic system to communicate in text. When a human were driving the system, the keys wouldn't light up; they'd just type freely.
Just to add clarification points, the term optogenetics is generally supposed to refer to any gene transfection that involves optical access, but in practice there are two separate categories:
* optogenetic stimulation: using light-activated ion channels to modulate neural activity
* optical imaging: using genetically targeted fluorescent proteins (mainly GCaMP) to observe the activity of neurons
This article is more the second kind, and the novel aspect is that this technology couples dopamine release to calcium rises, enabling imaging of the calcium signal to be used as a proxy for the neurotransmitter's release. It's more likely to be quite useful for investigating specific scientific questions about neurotransmitter signaling than a general purpose readout for BMI.
Multielectrode arrays pick up spikes or action potentials from individual neurons directly, so you get high temporal resolution but from a sparse sample of cells. This sample turns out to be enough to do a lot of interesting things, such as controlling a mouse cursor or a robotic arm. This is already in clinical trials at a few sites, including (my lab at) Stanford.
Imaging approaches are really powerful, but the signals are often slower. This has to do with the kinetics of the proteins themselves and the signal that is being detected (calcium is slower than voltage transients). But you get to see the activity of lots of neurons. There are voltage sensitive channels that you can image, although the signal to noise ratio isn't nearly as high as yet. It's not immediately clear how this could be used for a BMI in humans, mainly because you would absolutely need optical access to the brain to image, so you'd either be opening up a window or implanting some kind of imaging sensor. The less invasive approaches you were hinting at are mostly in the first category (stimulation), where for longer wavelengths of light you wouldn't need something as invasive.
Graduate students are students. > Yes, they are. This case is not about that, as it is not in dispute.
I do actually think this is disputable. As a late stage PhD candidate, I'm not sure that what I do for a living looks anything like "studying". I and my colleagues work on research problems with variable amounts of input from our advisors. We typically don't take classes after 2 years. I work 9-9 6 days a week on building experimental setups and tackling scientific questions that will be beneficial to both my employer (more papers + more grants = more overhead money for the University and more research funding for my advisor) and to my career. Yes, academia doesn't have a commercial output, but what I do isn't that dissimilar to what being a data/research scientist at a small company would do.
People (successfully) pursuing a PhD are almost by definition not "studying" anything; they're advancing the state of the art in their field. To me calling them students seems like a mechanism to depress salaries.
Definitely think it would be interesting to see YC Research's take on economics research, but I'm left wondering, why would someone want to take this position? As opposed to, say, entering an economics PhD program and studying the problem there. To be clear, I don't mean to spark an academia-vs-industry (or whatever YC research would be categorized as) battle. I don't know the field of econ research well and my prior would be that more perspectives and approaches to the issue would be helpful in studying the problem. What I'm interested in what would someone personally want out of doing this research but without the support of collaborators in a more formally collegial academic setting or the reputation and pull of an established think tank.
To be clear, I'm not trying to argue that taking this position is a bad idea, I'm just trying to better understand the opportunity and why YC research might be a better decision than a more run-of-the-mill PhD. I'm ever so slightly skeptical, but mostly just curious.
It's a great question, and I think this is a pretty good insight into the general state of the field of neuroscience at the cellular level. We know a lot of details about things going on inside neurons, and a lot of details about synapses. And these details are often specific to one of the hundreds of different types of neurons that are found in different parts of the brain. (e.g. http://www.neuroelectro.org/neuron/index/). And we do have methods to measure and manipulate various things in neurons in a dish. But the dynamics of a neuron's voltage are complicated, non-linear, and time-varying, and there are many parameters (e.g. the concentrations of numerous ionic species and other small molecules, many of which we probably don't even know about yet).
Even then, going from these messy biological details (e.g. these 20 proteins assemble into a particular form and release this neurotransmitter from this synapse when X happens) to an explanation for how the neuron works at a more algorithmic level is hard, and the field isn't there yet. Assembling and abstracting the details is hard and it's one of the goals of theoretical neuroscience. The complexity is probably a symptom of our lack of understanding, rather than the cause of it, i.e. there probably are a lot of details that we can abstract away in a simpler functional model.
I haven't read the paper, and I'm only vaguely familiar with Hawkins et al.'s HTM work. But I disagree with the claim at the end of the TR piece that these predictions are imminently testable. Thinking up a specific experiment to try and disprove theoretical ideas is often the hardest part of experimental neuroscience.
Paul was a close friend of mine, and we worked together on his research during his neurosurgery residency at Stanford. He was a hell of a human being, brilliant, dedicated, creative, refreshingly optimistic, and selfless. Before he was diagnosed, I'd planned to postdoc with him when he was considering starting a functional neurosurgery lab (actively manipulating the nervous system to achieve therapeutic benefit, with deep brain stimulation being the most successful example).
This piece he wrote shortly before he died is well worth your time. http://stanmed.stanford.edu/2015spring/before-i-go.html
This seems to follow Betteridges' Law. No coherent argument made for the case that there's any interaction between tech and jewelry. Intuitively, I think the Millenial generation cares less about ways to show material wealth, so that extends to cars, jewelry, clothing, etc...
Read the article, but I'm not sure what point you're trying to make with it. Of course I don't think that scientists "should" make less money as they're doing something they love, but since when are salaries determined by how much someone "deserves" to make? My working conditions are excellent; I'm at Stanford in a well-funded lab. And I decided and continue to believe that this is research that I'd like to do despite the financial opportunity cost.
If I'm reading it correctly, this article seems to suggest that the advice that pursuing work that you find fulfilling is itself responsible for lowering wages, which may be true in as much as people are willing to accept lower wages if they enjoy the work despite having better options. But what would you propose as an alternative? Pursuing work that you find less fulfilling in exchange for more money? That's not a sacrifice I'm willing to make, and I've been privileged with enough opportunity and financial security to have a choice. Of course everyone has different objective functions they're seeking to optimize, so I see why others would make different choices, which is all good by me.
Certainly a depressing perspective, but as a Stanford PhD student (neuro), my thoughts have always been that pursuing science was a decision to work on the problems that interested me at the _expense_ of not receiving good financial compensation. The particular things I'm interested in studying happen to exist primarily within academia (and non-university academic institutions like Allen Brain and Janelia), because the neuroscience work being done in industry (today) is far more primitive (e.g. EEG). This may change in the near future, and I'll reconsider my options then, but for now, I'm under no delusion that my salary (~30k) is anywhere near what it could be for an EE/CS in industry. That being said, if the amount of bullshit and politics becomes so burdensome that it kills the attractiveness of the science, then I'd leave.
Sometimes results won't replicate because the people or equipment at the second lab aren't skilled enough to do the experiment correctly. You can get type II errors on difficult tests because of impurities, noise, etc. But most of the tests listed on Science Exchange seem like standard tests, where the labs doing it are probably better at it than you are, so this wouldn't be too much of an issue.
I think another effect of this experiment-for-fee model would be to level the playing field for smaller labs that can't master every tool simultaneously. Whereas before a researcher would have to choose between (a) learning a new technique and buying new equipment and (b) finding an alternative way to demonstrate a result, presumably with older or less reliable methods, now they can do a very simple one-off collaboration of sorts without any of the overhead of real collaborations (esp. negotiations over authorship).
I managed to get this building on MacOS Mavericks using the glew and glfw3 libraries off of homebrew. I forked it at djoshea/imgui
See https://github.com/djoshea/imgui/blob/master/examples/opengl...
brew install glew
brew install glfw3
cd examples/opengl_example/
make
./main