Analytics Vidhya

This blog post has a nice visualization of sudoku solving using backtracking. https://medium.com/analytics-vidhya/sudoku-backtracking-algorithm-and-… Although the case they use is almost perversely simple. Imagine in a 'real' sudoku that each number might need to be backtracked multiple times, such that the algorithm is regularly going almost back to the start. Edit: here's a classic one trying to solve the 8 queens problem: https://commons.wikimedia.org/wiki/File:Eight-queens-animation.gif

Hi all, This was a fun project exploring the Raspberry Pi Zero W hardware, Go, modern JS APIs, and 3D Printing. I wanted to be able to get highlights of activity from my birdfeeder, and eventually run an onboard image classifier [0] to figure out which one of my neighbors have not been cleaning up after their dog. Compared to most other RPi motion detection projects, this runs more efficently by streaming H.264 video instead of MJPEG, and detection is based on motion vectors instead of image difference techniques. After testing a few other projects in this space, I'd recommend taking a look at PiKrellCam [1] if you really need to pinch CPU cycles on a RPi. For reference PiKrellCam runs at least twice as fast as Sentry-Picam, and has even better motion detection. [0] https://medium.com/analytics-vidhya/towards-gpu-accelerated-image-clas… [1] http://billw2.github.io/pikrellcam/pikrellcam.html

> The metric is also very human friendly: think of it as 30% of players in the given position, at the given game’s stage and state, successfully continued on to a victory What I am trying to say is that this is a meaningless piece of information, because that's not how games work. You can't freeze the game state and say "I have a xx% chance to win here", because the state of the game continues evolving over time. If you were actually estimating your chance to win, it would be a forecast. Except if it's anything like this kind of NN [0], then it is literally just estimating based on a snapshot of game state and not the context of the game. At best, this "win probability" is snapshot estimate of the current game state and has nothing to do with the actual outcome of the game. This value is also impossible to falsify, so it could be spitting out any random number and you can't say that it's wrong. [0] https://medium.com/analytics-vidhya/a-simple-neural-network-for-in-gam…

I was going to provide my passwords to any random person on the internet, Troy Hunt might be close to the top of the list, but I think your sentiment is sensible. I remember searching the dataset being fairly straight forward. It's been a while since I've done it, but I think I just downloaded the text file and then grepped it for hashes of my passwords, but I see people doing much more useful things: https://medium.com/analytics-vidhya/creating-a-local-version-of-the-ha…

I had a blog on it on overview of it with codes and explanation (more like eli15): https://medium.com/analytics-vidhya/googles-t5-transformer-theory-ffd0…

Good overview of the two approaches: https://medium.com/analytics-vidhya/why-are-lstms-struggling-to-matchu…

This is a pretty good introductory primer. https://medium.com/analytics-vidhya/understanding-the-mel-spectrogram-… 1. STFT (get frequencies from the audio signal) 2. Log scale/ decibel scale (since we hear on the log scale) 3. Optionally convert to the Mel scale (filters to how humans hear) Happy to answer any questions

You can already do this today with standard toolchain. https://medium.com/analytics-vidhya/running-go-code-from-python-a65b3a…

Read about "RNN Vanishing Gradients". LSTMs help here, but see eg https://medium.com/analytics-vidhya/why-are-lstms-struggling-to-matchu… for the problems there.

It seems like he was one of the people advocating using CUDA for deep learning before Krizhevsky’s big breakthrough in 2012. Not a primary source: https://medium.com/analytics-vidhya/gpu-for-deep-learning-7f4ef099b702 Alex Krizhevsky’s home page has a bunch of CUDA ConvNet stuff from ~2010 https://www.cs.toronto.edu/~kriz/ Edit: found a better source. Deep learning on GPUs dates to 2005, and Ng’s group has the first GPGPU ConvNet report in 2009, and Krizhevsky has the breakthrough in 2011/2 https://www.deeplearningbook.org/contents/applications.html

Here's a tutorial : https://medium.com/analytics-vidhya/install-tensorflow-2-for-amd-gpus-… I haven't tested myself, but since I have PTSD from Linux CUDA installations I'm glad competition is finally here.

You are correct New results show it's 5x slower on the M1 https://medium.com/analytics-vidhya/m1-mac-mini-scores-higher-than-my-… Interestingly, it's only utilizing the neural engine and no GPU. It'd be interesting to see what happens when M1 Tensorflow updates to utilize both GPU + NE

This is the differential equations paper I was thinking about https://medium.com/analytics-vidhya/solving-differential-equations-wit…

Curious what analysis you're referencing here. While I don't doubt people have shown that various transformer models have certain limitations, I'm pretty bullish on transformer models in general. Here's a post exploring the application of transformers to symbolic mathematics for instance: https://medium.com/analytics-vidhya/solving-differential-equations-wit…

See also: Erlang's crash on error, with clean restart philosophy. https://medium.com/analytics-vidhya/erlang-is-not-just-a-language-it-i…

Basically instead of FROM python:3.9, you do FROM scratch. This represents an empty container with nothing. It's quite easy to use with statically compiled languages, and a bit less so with interpreted ones like Python, but you need to use a multi-step build with first an image containing your dependencies to compile, and then a stage from scratch copying just the resulting artifacts. That way your production image is extremely light and has no attack surface or stuff to update outside of your own code. The main downside is that debugging live is not easy (if you have some networking errors, you can't just exec inside and do some ping/curl/netstat/etc. to check stuff, because you can't even exec inside due to no shell being present), and requires extra tooling like cntr/cdebug or Kubernetes Ephemeral Containers which will attach a fuller container (e.g. busybox) to the running scratch one to be able to debug. https://chemidy.medium.com/create-the-smallest-and-secured-golang-dock… https://medium.com/analytics-vidhya/dockerizing-a-rest-api-in-python-l…

If you can’t fit the model on your resources you can leverage DeepSpeed’s ZeRO-offload which will let you train GPT2 on a single V100 (32gb). Alternatively, if you’re researching (with the caveat that you have to either publish, open source or share your results in a blog post) you can also get access to Google’s TPU research cloud which gives you a few v3-8s for 30 days (can’t do distributed training on devices but can run workloads in parallel). You can also ask nicely for a pod, I’ve been granted access to a v3-32 for 14 days pretty trivially which (if optimized) has more throughput than 8xA100 on transformer models. TPUs and moreso pods are a bit harder to work with and TF performs far better than PyTorch on them. https://www.deepspeed.ai/tutorials/zero-offload/ https://medium.com/analytics-vidhya/googles-tpu-research-cloud-free-tp…

https://sites.google.com/view/avhyas-sac-isro/home https://github.com/AVHYAS-QGIS-Plugin/version-1 https://arxiv.org/pdf/2106.12776 https://medium.com/analytics-vidhya/avhyas-a-python-based-qgis-plugin-… https://ieeexplore.ieee.org/document/9574057

>"...Graphics programming has this intriguing concept of 4D vectors used to represent 3D objects..." Intriguing indeed! Related: https://www.popularmechanics.com/science/math/a27737/visualize-higher-… https://www.quora.com/What-fields-of-mathematics-deal-with-higher-dime… https://medium.com/analytics-vidhya/understanding-the-mathematics-of-h… https://piggsboson.medium.com/higher-dimensions-for-dummies-147011dc30… https://www.cs.cmu.edu/~venkatg/teaching/CStheory-infoage/chap1-high-d… https://encyclopediaofmath.org/wiki/Higher-dimensional_geometry https://yashkant.github.io/spad/ https://en.wikipedia.org/wiki/Pl%C3%BCcker_embedding https://en.wikipedia.org/wiki/Grassmannian