Nalin Dhiman

विचार है।

[Nalin Dhiman @dhiman_nalin]

हज़ारों ख्वाहिशें जिंदा जलाती हैं, ये ज़िंदगी यूं है की चलती ही जाती है, बारिश में भी पानी को तरसाती है, ये ज़िंदगी यूं है की चलती ही जाती है,रण में धूप को तरसाती है, ये ज़िंदगी यूं ही चलती जाती है, खुद के नाम पे बहरूपिए से मिलाती है , ये ज़िंदगी यूं ही चलती जाती है।

[Nalin Dhiman @dhiman_nalin]

It is an amazing feat of software engineering to bridge these two systems, but marketing it as a purely biologically driven "conscious" upload walking around is a massive exaggeration of what the math is actually doing.

[Nalin Dhiman @dhiman_nalin]

The Reality of the "Whole Brain Upload" in a Nutshell: Eon Systems did not put a biological mind inside a robot and watch it teach itself to walk they built an incredibly complex organic digital puppet. Because the FlyWire connectome only maps the upper brain

[Dr. Alex Wissner-Gross @alexwg]

https://t.co/N2AwpI9f8m

[Nalin Dhiman @dhiman_nalin]

The brain isn't driving the body; the brain is just pressing a button on a remote control, and a preprogrammed robot script is executing the walking animation.

[Nalin Dhiman @dhiman_nalin]

@SagasofBharat I'd love to see you derive the fundamental theorem of calculus using only Aryabhatas shlokas. you’d be googling Newton by line 3. Stuck romanticising old verses while typing on a device built on Newtons calculus, adorable levels of delusion you people have. cute!

[Nalin Dhiman @dhiman_nalin]

@SagasofBharat Impressive precursors, sure, but without Newton's limits, derivatives, and integrals (built on Leibniz's notation we still use), youd have no rockets, no GPS, no AI memes. It's like crediting the wheel for your tesla. you bullock cart people wont understand

[Nalin Dhiman @dhiman_nalin]

@michaelandregg @FlyWireNews @Philip_Shiu Is this feedback loop biologically isomorphic, or an artificial heuristic designed to keep the CPG controller balanced?

[Nalin Dhiman @dhiman_nalin]

@michaelandregg @FlyWireNews @Philip_Shiu Walking requires immense high-frequency closed-loop sensory feedback. In your MuJoCo environment, how exactly are you mathematically converting the digital

[Nalin Dhiman @dhiman_nalin]

@michaelandregg @FlyWireNews @Philip_Shiu floating-point joint collision data back into biologically meaningful spike trains injected into the fly's proprioceptive sensory neurons?

[Nalin Dhiman @dhiman_nalin]

@michaelandregg @FlyWireNews @Philip_Shiu Did you have to selectively mute neurons or introduce mean-field approximations to prevent the 130k-node LIF network from collapsing into chaotic noise or seizing, or is it truly 1:1 spiking without artificial dampening?

[Nalin Dhiman @dhiman_nalin]

@michaelandregg @FlyWireNews @Philip_Shiu You achieved this using a widespread LIF model. However, Drosophila neurons rely heavily on graded (non-spiking) transmission and gap junctions.

[Nalin Dhiman @dhiman_nalin]

@michaelandregg @FlyWireNews @Philip_Shiu Translating discrete LIF brain spikes into the continuous MuJoCo torque required to achieve '95% behavioural accuracy' seems mathematically daunting without Reinforcement Learning!

[Nalin Dhiman @dhiman_nalin]

@michaelandregg @FlyWireNews @Philip_Shiu Since biological muscles use soft-tissue hydraulics and MuJoCo uses rigid-body digital actuators, how heavily did your team have to manually tune the mapping transfer functions?

[Nalin Dhiman @dhiman_nalin]

@michaelandregg @FlyWireNews @Philip_Shiu How much of this autonomous MuJoCo walking behaviour is driven purely by the brain connectome, versus the human-coded hybrid CPG (spinal/VNC) controller stepping the legs in rhythm?

[Nalin Dhiman @dhiman_nalin]

@michaelandregg @FlyWireNews @Philip_Shiu My question is: the FlyWire v783 dataset only maps down to Descending Neurons (DNs). The NeuroMechFly v2 paper mentions using a 'hybrid low-level motor controller integrating CPGs' for movement

[Nalin Dhiman @dhiman_nalin]

@alexwg Is this feedback loop biologically isomorphic, or an artificial heuristic designed to keep the CPG controller balanced?

[Nalin Dhiman @dhiman_nalin]

@alexwg Walking requiresimmense highfrequency closedloop sensoryfeedback inyour MuJoCo environment how exactly are you mathematically converting the digital floatingpoint joint collision databack intobiologically meaningful spike trains injected into the flys proprioceptive sensory neuro