vervez

There's also recent Bayesian Optimal Experimental Design methods that allow you to directly design experiments using gradient ascent. Not sure how it compares with BayesOpt on your problem, though.

github copilot won't pay for itself.

Their paper is more for the case of "we can't gather more data, so what to do?" but your solution is in line with optimal experimental design and choosing a utility function to distinguish between models using as little data as possible.

Here's a good recent paper that looks at this problem and provides remedies in a Bayesian manner. https://arxiv.org/abs/2202.11678

min{d(problem)/d(theta)} is essentially what LLMs are doing with a prompt. Every session that a chatgpt user has leads to either a resolution or not, which is the loss function given the prompt used to reach that point. It's getting better at not hallucinating in my experience over just the past 4 months.

Thanks for sharing and i'm happy to be your friend :)

I see your point and agree it can be frustrating if DEI initiatives are used as a political tool or blindly without considering the pitfalls of a naive implementation.

Indeed, I didn't cite any sources but I'm fairly sure there's literature studying this phenomenon. I agree that DEI when used as a political tool is a distraction for a company.

I think many of the things you mentioned w.r.t. diversity is actually forward-thinking and potentially yields more returns later. If there's a brilliant swe that feels uncomfortable at a company because of how they name their branches, they won't work there. Same goes for hiring a "less-qualified" individual. How many decisions or projects can be improved because of a diversity of input? Again, a non-linear outcome that is harder to measure than immediate profitability. Those are some arguments for those programs focusing on your productivity point.

Equivariance is growing in popularity in machine learning, so these tricks will be helpful if one wants to study, or publish, in that area (I'm thinking about it). I'd recommend for any ML-related folks to look into this area and save this thread whenever they're trying to implement an equivariant neural network.

So you `pip install`ed them at a new company? I'll see myself out.

Here's a short (unscientific) article (that kinda links to scinetific articles) about the link between lactic acid buildup and its relation in the brain. Sounds like exercise generates a natural nootropic. https://www.outsideonline.com/outdoor-adventure/lactic-acid-...

Very cool. Can you comment on how this compares with tensorflow similarity? https://blog.tensorflow.org/2021/09/introducing-tensorflow-s...

Couldn't read the article but yea, if it's a small molecule, most likely it's inhibiting some protein specific to cancerous cells. In this case, it sounds like it's blocking some protein that blocks human cells' innate ability to produce antigens, which signal to T-cells that they are defective and need to be destroyed.

Sometimes we understand the biology after we discover a treatment.

Briefly, no. Why: they're most likely not training your average ML models on this dataset. Instead, they are likely taking a model of some physics and seeing how it performs in these simulations.

You can think of this as a form of causal inference - "if this model is true, how well does it work with our current understanding (simulations) of physics?" type of questions. There are measures of error and bias that come with evaluation of these models.

That's a strategy but there are issues with immune rejection since that's a xenograft. I think there are some products based off of this concept for other areas (skin grafts) after removing all xeno-cells from the tissue, but articular cartilage is more difficult to remove/replace xeno-cells with autologous or allogenic cells.

Bioengineer here able to chime in with niche expertise. Cartilage is and has been a holy grail in biomedical engineering but is very difficult to grow and transplant. There have been some successful neo-cartilage projects but integrating that into a defect and successfully integrating with host cartilage is the problem. The boundary of the defect actually has electronegative components that actively oppose integration of host and transplant (synthetic) cartilage. I think the most likely solution won't be a tissue-engineered transplant but rather an active cellular component that appropriately responds, and builds upon, this negative feedback cycle via cellular programming.

The way I've seen startups make it work is to save part of the indirect for the worst-case scenario of no bridge funding for the 3-6 month period. If you are able to get into an accelerator that takes a small cut and aligns with your area, the amount of indirect saved (.4*250k) available could be sufficient to bridge a low-cost org for 3-6 months. Even then, it's still a gamble as to whether you'll receive the Phase II in that time.

As for timeline, you're able to apply for the Phase II to kick in right as the Phase I is ending. I've seen that work but it requires planning and long hours to perform research and write the next phase proposal. There's also direct to Phase II for ~$2m in one grant.

If you have outside responsibilities and don't want to risk that bridge, you could try and submit the direct to Phase II, which would also help develop your idea to pitch to VCs in clean-tech space.

I'd be interested in seeing citations for the known death valley. This is the second time I've heard this anecdote but have never seen the evidence.

With hard science ventures that sometimes have a hard time findng the 'killer app' for the technology, I think the experience and learning from ping ponging around is a feature. Yes, startups need to plan for the long and uncertain apply-decide-reapply cycle

SBIR funding is critical for hard science ventures early on. It's also a good indicator to future investors of potential hard science projects that a panel of experts in the area has reviewed and approved government funding for the idea, and the team is at least decently competent to meet the milestones of the SBIR. This helped Ginkgo Bioworks before they received more than half a billion in private investment.

I'm not an economist but this also seems like an indirect, peer to peer, way to establish 'credit' of an individual, rather than relying on centralized institutions with dubious security practices, like Equifax. Is this obvious or has someone already thought of this and I'm being dumb? I don't typically hear about this type of application of digital currencies in mainstream media but it sounds interesting in certain applications of peer-determined credit.

Is Morgan Freeman the most used celebrity?

This could work out pretty well. I remember reading somewhere (can't find the article) that Brian Eno works with artists this way as a producer, creating artificial constraints to expand their creativity. Would be interesting to see how this affects the startup process where optionality can be overwhelming.

Another possibility: the bacteria could be engineered to switch to a different metabolic mechansim, like the use of the lac operon to digest lactose instead of glucose whenever there's a scarcity of glucose. Although this envisages stably engineered bacteria that switch between plastic and regular metabolic process, which I'm not sure about the current feasibility/precision.

You're not wrong, I'm just saying they're synonymous. https://www.hsmagnets.com/blog/halbach-array/

Halbach array AKA fridge magnet.

Scapegoat?

Maybe because the lifetime of the computer will be shorter, hence, closer to the next sale?

Excellent post about potential flaws of your typical linear regression model and how to address with causal methods. As a self-taught practitioner, these posts are priceless in letting me know what I don't know...