Venkat Ramaswamy

Lindsay Lab - Postdoc Position Artificial neural networks applied to psychology, neuroscience, and climate change

Spread the word: I'm looking to hire a postdoc to explore the concept of attention (as studied in psych/neuro, not the transformer mechanism) in large Vision-Language Models. More details here: lindsay-lab.github.io/2025/12/08/p...
#MLSky #neurojobs #compneuro

[moments after creating my AI digital clone]

ME: alright clone, do my chores

CLONE-ME: no

US: my god, it worked

Here at #Indian academy of #neuroscience meeting in the beautiful city of Kovalam in Kerala share.google/EzLqUeTwCeDw... ♥️

@sbaulac.bsky.social is giving the opening talk on brain mosaicism in epilepsy and cortical malformations #epilepsy.

Excited to have 2 papers accepted at @neuripsconf.bsky.social 2025 - Reliable ML Workshop.

Paper #1: openreview.net/forum?id=9Ue...
Paper #2: openreview.net/forum?id=0MW...

I’ll briefly summarize the work below.
(Links above if you want to get right to the papers)

THREAD 🧵

Please RT.
(1/N)

Thanks Krishna!

a little girl is sliding down an orange slide with the words ok bye written on it . ALT: a little girl is sliding down an orange slide with the words ok bye written on it .

I'll stop here.

Both these papers were led by
@simran-ketha.bsky.social
, an extraordinary Ph.D. student advised by me.

Watch out for many new (and yes exciting!) results that we hope to share soon.
(N/N)
\END

We asked if we could similarly leverage the subspace geometry to obtain robustness to adversarial attacks. We built variants of MASC that ended up being upto ~3x better than the model, even though both use the same initial substrate.

More here: openreview.net/pdf?id=0MWW5...

(10/N)

a cartoon of a monkey sitting on a rock meditating ALT: a cartoon of a monkey sitting on a rock meditating

Defending against adversarial attacks has become important. Effective defenses usually involve adversarial training, which is expensive, or other ways to defend that involve modifications to standard training paradigms.

But, what if the key to defending from adversarial attacks lies within?
<9/N>

So, what’s an adversarial attack?

It turns out that with typical Deep Networks a malicious adversary can change inputs, e.g. images so the Deep Net classifies it as something else. See the example image below from blog.mi.hdm-stuttgart.de/index.php/20...

(8/N)

Paper #2 w/
@simran-ketha.bsky.social
@mummani-nuthan.bsky.social
&
@niranjanrajesh.bsky.social

Here we considered the setting of adversarial attacks.

(7/N)

We don’t know why this works well.

Indeed, it is reminiscent of some Neuroscience experiments, where it is known that animals sometimes have significantly poorer behavioral performance than what one can linearly decode from a handful of their neurons.

Paper: openreview.net/pdf?id=9Uen9...

(6/N)

Surprisingly, we find that this works extraordinarily well.
For every model tested, on at least one layer MASC beats the model test accuracy, and in many cases by a significant margin (see table below), and especially for cases where there is high degree of corruption of labels.
(5/N)

More technically, we fitted subspaces, one for each class, to the *corrupted* layerwise outputs. For each incoming point, we asked, which subspace is it closest to, in the angle sense, & predicted the label of the datapoint to be that of that class. We call this the MASC classifier.

(4/N)

We looked into the internals of Deep Networks to see if we could extract much better generalization (i.e. accuracy on unseen data)

Specifically, we looked at the geometry of class-wise internal representations and whether it was organized in a manner that allowed for better generalization.

(3/N)

a group of children are sitting at their desks in a classroom covering their faces with their hands . ALT: a group of children are sitting at their desks in a classroom covering their faces with their hands .

Paper #1 w/
@simran-ketha.bsky.social

We consider the setting, where the training data has label noise. That is, w/ some probability, each training point has its label shuffled.

Here, Deep Nets *memorize*, i.e. are able to perfectly rote-learn the training data but do badly on unseen data
(2/N)

Excited to have 2 papers accepted at @neuripsconf.bsky.social 2025 - Reliable ML Workshop.

Paper #1: openreview.net/forum?id=9Ue...
Paper #2: openreview.net/forum?id=0MW...

I’ll briefly summarize the work below.
(Links above if you want to get right to the papers)

THREAD 🧵

Please RT.
(1/N)

Mice navigate scent trails using predictive policies Animals actively sense their environment to extract features of interest to guide behaviors. For mammals, odors are prominent environmental features which are sampled by active modulation of sniffing ...

Thrilled to share this work, long time in the making! Carried through creatively by @siddjakes.bsky.social after initial design & piloting by @trackingskills.bsky.social, with help from @trackingactions.bsky.social. Modeling in collaboration with Massimo Vergassola & Nicola Rigolli.

www.thewayofcode.com by
@rickrubin.bsky.social

"Early to bed, early to rise, work like hell, and advertise."
-Ted Turner

Ian, how about a bit of intellectual honesty in posting the full official statement by India, so your readers can make their own judgement about the rationale for India's position?

Will start my posts here with a preprint!
First preprint from my postdoctoral work—where we redefine and remap the isocortical efferent projectome through two foundational neurogenic mechanisms.
*
(1/5)

Our new paper out now in Science explores how neural activity in the lateral entorhinal cortex (LEC) *drifts* over time - and *jumps* at key boundaries - to help organize events in memory.

🔗 www.science.org/doi/10.1126/...

Here's a quick summary of what we found 🧵👇

(1/7) New preprint from Rajan lab! 🧠🤖
@ryanpaulbadman1.bsky.social & Riley Simmons-Edler show–through cog sci, neuro & ethology–how an AI agent with fewer ‘neurons’ than an insect can forage, find safety & dodge predators in a virtual world. Here's what we built

Preprint: arxiv.org/pdf/2506.06981

Foundations of Computer Vision The print version was published by

Our computer vision textbook is now available for free online here:
visionbook.mit.edu

We are working on adding some interactive components like search and (beta) integration with LLMs.

Hope this is useful and feel free to submit Github issues to help us improve the text!

Exendin-4, the first GLP1 receptor agonist on market came from Gila Monster's venom that contains a mammalian GLP1 homolog. Without it we wouldn’t have had the current medical revolution that Ozempic etc. wrought.

Almost all Nobel Prizes are awarded for work that is exploratory, or absolutely basic science with no obvious commercial or medical benefit.

You cannot predict where advancements come from, so you have to invest in science and scientists.

Targeted (corporate) science investment will never do this.

Memory by a thousand rules: Automated discovery of functional multi-type plasticity rules reveals variety &amp; degeneracy at the heart of learning Synaptic plasticity is the basis of learning and memory, but the link between synaptic changes and neural function remains elusive. Here, we used automated search algorithms to obtain thousands of str...

We just pushed “Memory by a 1000 rules” onto bioRxiv, where we use clever #ML to find #plasticity quadruplets (EE, EI, IE, II) that learn basic stability in spiking nets. Why is it cool? We find 1000s!! of solutions, and they don’t just stabilise. They #memorise! www.biorxiv.org/content/10.1...

Subcortical correlates of consciousness with human single neuron recordings Subcortical brain structures contain neurons encoding subjective reports of stimulus detection in human participants, suggesting a role for conscious perception.

New “version of record” in @elife.bsky.social! ‬

We recorded neurons in the thalamus and subthalamic nucleus of humans and found both sensory- and perception-selective neurons with two distinct latencies!

7-year project with @nfaivre.bsky.social + @foscobernasconi.bsky.social.

A short thread 👇

Conserved brain-wide emergence of emotional response from sensory experience in humans and mice Emotional responses to sensory experience are central to the human condition in health and disease. We hypothesized that principles governing the emergence of emotion from sensation might be discovera...

This! This the type of work that makes me so optimistic that the next 3 decades of brain research will be more impactful that the last for understanding/treating mental conditions. We finally have the right measurement tools & analyses (eg for emotion).

www.science.org/doi/10.1126/...

#neuroskyence What are your favorite papers on the idea of redundancy in the brain? E.g. how a different neural circuit can step in when the "original" is damaged? Or just how the same function can be executed in different ways within the same brain?