Ashutosh
banner
conscioustahoe.bsky.social
Ashutosh
@conscioustahoe.bsky.social
researcher @ genzel lab; teaching assistant @ neuromatch; computational cognitive science and behavioural science; language, memory and mental health; https://conscioustahoe.github.io
Merry Christmas everyone 🎄
December 25, 2023 at 6:39 PM
Day 21: Read 2 studies on neural correlates of memory, specifically focussing knowledge retrieval and updating knowledge. #neuroscience #deeplearning
December 25, 2023 at 6:37 PM
Day 20: Crossed a major hurdle today. Now, my sleep model can tell the accuracy of wakefulness, REM and NREM sleep using EEG+EMG signals. #100DaysOfCode #neuroscience #deeplearning
December 24, 2023 at 10:18 PM
Day 19: Today I read 2 blog posts on Understanding GPU Memory by Pytorch
Read them here t.co/v4NC15dWrq and here t.co/o9dNDsMJhI. I love how simple and direct they are written, no bs. #100DaysOfCode #DeepLearning
Understanding GPU Memory 1: Visualizing All Allocations over Time
During your time with PyTorch on GPUs, you may be familiar with this common error message:
t.co
December 20, 2023 at 8:06 PM
Day 18: I was consumed by a jax issue #100DaysOfCode
December 19, 2023 at 8:49 PM
Day 17: More than month since I posted my progress. Been busy with applications, job and lab work! Mostly worked on pipeline to infer and analyse latent sleep states. Hoping to make more progress before I go on vacation next week #100DaysOfCode #deeplearning
December 18, 2023 at 6:27 PM
Day 16: Reviewed matrix calculus (gradients, jacobiens, scalar and jacobian chain rule). Also worked on improving brain2image retrievals using SigLIP, now have decent performance. #100DaysOfCode #neuroimaging #neuroscience
October 29, 2023 at 7:03 PM
Day 15: Theory of extracellular fields and currents, EEG recording from the scalp, ECoG recorded by the subdural grid electrodes from the cortical surface and LFP and contributors of extracellular fields(ionic processes from fast action potentials to slowest fluctuations)#100DaysOfCode #neuroskyence
October 23, 2023 at 7:46 PM
Day 14: Learned new things related to signal processing; spent a lot of time inspecting hpc and pfc data to understand why the 3d plot was not making sense; turns out you need to look at it from the correct angle (feels silly). Also learned about fft and pwelch analysis. #100DaysOfCode #neuroskyence
October 8, 2023 at 8:09 PM
Day 13: Attempted to improve the brain2image forward and backward retrievals by making the labels soft in the SigLIP loss function and somehow made it worse. I will continue tomorrow! #100DaysOfCode #neuroskyence
October 3, 2023 at 9:07 PM
Day 12: Trained mcRBM model again after minor changes and increasing the batch size from 256 to 16384 on an A4000; took 7 mins to run 10K epochs. #100DaysOfCode #neuroskyence
October 1, 2023 at 10:56 PM
Day 11: Trained mcRBM model on the sleep scoring dataset. mcRBM is an energy-based model used for unsupervised learning. #100DaysOfCode #neuroskyence
September 30, 2023 at 9:04 PM
Day 10: In the MedARC journal club, today we discussed the BrainLM foundation fMRI model introduced by Van Dijk Lab.

Preprint: t.co/MUobqXULfb

#100DaysOfCode #neuroskyence
BrainLM: A foundation model for brain activity recordings
bioRxiv - the preprint server for biology, operated by Cold Spring Harbor Laboratory, a research and educational institution
t.co
September 29, 2023 at 8:06 PM
Day 9: Convolutions on manifolds! Okay, in image convolutions we have a grid of numbers (the filter) that we slide over an image to perform operations like edge detection. On a manifold, we need to adapt this because there's no natural grid structure. #100DayOfCode #AcademicSky
September 26, 2023 at 6:51 PM
Day 8: The concept of manifolds is essential to how diffusion models work and I never really understood them before and after spending a few hours today I only understand them to a small extent. Maybe a few more days! Let's see what I know now #100DayOfCode #AcademicSky
September 25, 2023 at 8:37 PM
Day 7: Made a lot of progress on SigLIP loss function I have been working on. I am able to see some decent performance metrics on brain2image retrievals and vice-versa (Test Bwd Pct Correct: 0.6875 Test Fwd Pct Correct: 0.8906). #100DaysOfCode #neuroscience #neuroskyence #deeplearning
September 24, 2023 at 7:09 PM
Found this today, "A Beginner's Guide to Neural Mechanisms"
VideoLectures
We are excited to share  A Beginner’s Guide To Neural Mechanisms , a free video-course entirely devoted to philosophy of neuroscience and neurophilosophy. Funding for this Beginner’s Guide was....
www.neuralmechanisms.org
September 23, 2023 at 9:33 PM
Day 6: Read a lot of open clip code to understand how CLIP loss is calculated for image and text vectors. Still more to read. Going to use this knowledge to implement a new loss function I am working on for brain2natural image retrievals. #100DaysOfCode #deeplearning
September 23, 2023 at 9:08 PM
Day 5: Interpolated EMG-Like signal to FFT time points and used it to creat 3-D plots with theta hippocampal and delta prefrontal cortex signals to visualize where energy of different sleep states(wakefulness, rem, nrem) rest. #100DaysOfCode #neuroscience #neuroskyence
September 22, 2023 at 6:59 PM
Day 4: Compared results of preprocessing after motion & slice time correction and smoothing with original data. I am wondering instead of Gaussian Kernels can we use VAE to learn compact representation of high-dimensional fMRI data while preserving important spatial patterns &features.#neuroskyence
September 21, 2023 at 8:39 PM
Day 3: Today I did brain image extraction from anatomical brain image and preprocessing steps on the functional brain data. #100DaysOfCode #neuroscience #neuroskyence #neuroimaging
September 20, 2023 at 6:15 PM
Day 2: Implemented SigLIP loss function for the train pipeline of MindEyeV2 (fMRI to natural scene images). The performance on brain2image and image2brain retrievals compared with soft clip loss are off at the moment! Need to work on that! #100DaysOfCode #neuroskyence #neuroimaging #deeplearning
September 19, 2023 at 3:12 PM
Day 1: Extracted beta, delta and theta band signals and EMG like signals from a sleep scoring dataset and later did simple time series analysis on the extracted signals.

#100DaysOfCode #neuroscience #neuroskyence
September 18, 2023 at 4:14 PM
I’m going to try to post my progress on a couple of projects in the space of computational neuroscience and deep learning here. I have never done this before, excited to see how I progress!

#100DaysOfCode #neuroscience #neuroskyence #deeplearning
September 18, 2023 at 4:04 PM