RocketSyntax

- Statistics

- Spiking neural nets

- Convolutional neural nets

- Paragliding

- Mountain biking

My mind went straight to gini impurity

I am planning to solve this =) just cracked the inner workings of ANNs (future SHOW HN) and am going to read a book on computational neuroscience tomorrow.

Reminds me of the book The Joy of UX

I thought AWS would win for sure due to its GovCloud adoption https://aws.amazon.com/govcloud-us/?whats-new-ess.sort-by=it...

You just re-fit the model on different data. It's called transfer learning. The theory is that it will run with similar features in your input data, therefore the weights will not need to be adjusted too much. You can also freeze the weights of entire layers.

No, it doesn't.

Need something to dominate networking at big conferences.

QR code?

I've heard this termed a Model Zoo. https://modelzoo.co/

This is why I love `for` loops, and why I hate functions that accept functions as arguments.

Sustainable code should be easy to read out loud.

1. Nameless icons in the bar of most visited pages.

2. Folders of topic/ project specific info pages.

3. 'Reading List' Chrome extension for one offs and things I want to come back to.

Biannual pruning of the folders.

You can use permuted feature importance to tell you what inputs are most correlated with the condition. WAYYYY better than the binary correlation methods most statistics use for each feature.

"The resulting algorithm outperformed all entrants at the most recent blind assessment of methods used to predict protein structures, generating the best structure for 25 out of 43 proteins, compared with 3 out of 43 for the next-best method."

"Cleaning up this pollution and paying for the public health and environmental damage costs the United States $157 billion annually."

Hot take - the Golden Circle underwhelms.

This inspires me even more to learn about boltzmann machines. The `why` lighting up the limbic brain

Ah, it's spiking edges as opposed to spiking node activations. I like this because it applies to 1 of many connections uniquely, rather than dividing the entire graph by spiking neurons. Exponentially more edges than nodes in a dense net. Connections are always more important than entities.

```The dendrites generated local spikes, had their own nonlinear input-output curves and had their own activation thresholds, distinct from those of the neuron as a whole```

How would this work in practice? Apply activations to multiplication values of the weight, or just don't perform the multiplication if the activation of the node is low?

Success is not just the hard work in the things that ultimately succeed.

It's more than meets the eye. It's also about the hard work in failures and decisions about things you decide not to do.

For example; trying 3 different workarounds to get 1 buggy piece of code to work.

At this point, why wouldn't you use Excel?

What use cases do you have that go beyond Excel?

I say this as someone that develops pipelines in notebooks and workflow frameworks.

Snowflake.

Being in the industry of identifying target variants from cohorts for pharma - I doubt 23andMe has a chemistry research organization dedicated to independently developing their own drugs.

Every time anyone does literally anything in genomics it gets publicized. That's how the industry works. https://www.genomeweb.com/

There is no substance to this.

Don't get me wrong, I think Microsoft is making all of the right moves. But AWS is releasing new services at breakneck pace.

MSFT has made a strategic investment in Databricks. This feels like where they can carve out more advantage.

MSFT acquired github... haven't seen much change here though. What if AWS acquires GitLab?

Probably at start of first exon and end of last exon. Or regulatory regions before them.

edit title to say "segments of genes"

Not soon for genes. Right now it is a painstaking process to learn what nucleotides do what because you need to sequence and analyze groups of people for each trait/ disease at scale.

We do biobank analysis w pharma to uncover drug targets for nasty diseases.

I wonder if reinforcement learning could tease out patterns faster.

Shiny

Knowing how to ask specific questions

- Differences between pytorch and keras.

- Spiking neural networks.

- Tensorboard

- Kubeflow

- Apache Beam.

- Knative.

- MLflow.

- Graph genomes.

- Traditional statistics.

No. I wouldn't do it. You can be a project manager right out of school. I was.

I'd recommend pushing for product manager/ product owner position. It's a pretty fluid role and many job reqs want computer science education.

https://www.linkedin.com/in/laynesadler/

Currently reading "The Master and his Emissary" basically says that right and left brain ensemble back and forth