pheonix1729's profile picture. Ph.D. Student, Indian Institute of Science.
Neuromorphic Engineering & Analog IC/FPGA Design | IEEE HKN ( #IamHKN) |   BITS Pilani Alumni

Kundan Kumar

@pheonix1729

Ph.D. Student, Indian Institute of Science. Neuromorphic Engineering & Analog IC/FPGA Design | IEEE HKN ( #IamHKN) | BITS Pilani Alumni

Kundan Kumar reposted

Many recent posts on free energy. Here is a summary from my class “Statistical mechanics of learning and computation” on the many relations between free energy, KL divergence, large deviation theory, entropy, Boltzmann distribution, cumulants, Legendre duality, saddle points,…

SuryaGanguli's tweet image. Many recent posts on free energy. Here is a summary from my class “Statistical mechanics of learning and computation” on the many relations between free energy, KL divergence, large deviation theory, entropy, Boltzmann distribution, cumulants, Legendre duality, saddle points,…

Kundan Kumar reposted

In continuous generative diffusion, the conditional entropy rate is the constant term that separates the score matching and the denoising score matching loss This can be directly interpreted as the information transfer (bit rate) from the state x_t and the final generation x_0.

LucaAmb's tweet image. In continuous generative diffusion, the conditional entropy rate is the constant term that separates the score matching and the denoising score matching loss

This can be directly interpreted as the information transfer (bit rate) from the state x_t and the final generation x_0.

Kundan Kumar reposted

Mathematicians are a bit weird sometimes

Anthony_Bonato's tweet image. Mathematicians are a bit weird sometimes

Kundan Kumar reposted

Jensen's inequality gives the difference between the average value of a convex function φ, and its value at the center, where both “average” and “center” are defined in terms of some distribution p_X. When the function φ is flat, or the distribution is narrow, they agree.


Kundan Kumar reposted

1/18 Today I will try to describe a mathematical trick that can logically explain what happens in Quantum Mechanics and Special Relativity.

matterasmachine's tweet image. 1/18
Today I will try to describe a mathematical trick that can logically explain what happens in Quantum Mechanics and Special Relativity.

Kundan Kumar reposted

On Tuesday, in my class, we have learnt that all a neural net does is stretching / contracting the space fabric. For example this 3-layer net (1 hidden layer of 100 positive neurons) gets its 5D logits (2D projections) linearly separable by the classifier hyperplanes (lines).


Kundan Kumar reposted

Happy birthday to Sir Andrew J. Wiles! He received the Abel Prize in 2016 for "his stunning proof of Fermat’s Last Theorem by way of the modularity conjecture for semistable elliptic curves, opening a new era in number theory." #Abelprize #Abelprize2016 #science #mathematics

abel_prize's tweet image. Happy birthday to Sir Andrew J. Wiles!

He received the Abel Prize in 2016 for "his stunning proof of Fermat’s Last Theorem by way of the modularity conjecture for semistable elliptic curves, opening a new era in number theory."

#Abelprize #Abelprize2016 #science #mathematics
abel_prize's tweet image. Happy birthday to Sir Andrew J. Wiles!

He received the Abel Prize in 2016 for "his stunning proof of Fermat’s Last Theorem by way of the modularity conjecture for semistable elliptic curves, opening a new era in number theory."

#Abelprize #Abelprize2016 #science #mathematics
abel_prize's tweet image. Happy birthday to Sir Andrew J. Wiles!

He received the Abel Prize in 2016 for "his stunning proof of Fermat’s Last Theorem by way of the modularity conjecture for semistable elliptic curves, opening a new era in number theory."

#Abelprize #Abelprize2016 #science #mathematics

Kundan Kumar reposted

The mathematician John Conway didn’t fit into a box. 🧵

QuantaMagazine's tweet image. The mathematician John Conway didn’t fit into a box.

🧵

Kundan Kumar reposted

Freeman Dyson on how he struggled to learn quantum mechanics from Paul Dirac


Kundan Kumar reposted

Reinforcement Learning

docmilanfar's tweet image. Reinforcement Learning

Kundan Kumar reposted

Oldies but goldies: Mark Kac, Can you hear the shape of a drum? 1966. Non-isometric shapes can have the same Laplacian eigenvalues (spectrum). Displaying here some Laplacian eigenvectors. en.wikipedia.org/wiki/Hearing_t…


Kundan Kumar reposted
vardi's tweet image.

Kundan Kumar reposted

Book #OTD "Functions of Matrices: Theory and Computation" by Nicholas J. Higham epubs.siam.org/doi/10.1137/1.…

FrnkNlsn's tweet image. Book #OTD 

"Functions of Matrices: Theory and Computation"
by Nicholas J. Higham

epubs.siam.org/doi/10.1137/1.…

Kundan Kumar reposted

The l^p functional is convex and hence a norm for p>=1. It is sparsity-inducing for p<=1. The l^1 norm is the heart of the lasso, aka basis pursuit. en.wikipedia.org/wiki/Basis_pur… en.wikipedia.org/wiki/Lasso_(st…


Kundan Kumar reposted

Exact Fisher-Rao geodesics and guaranteed approximations of Fisher-Rao distances for multivariate normal distributions. doi.org/10.1016/bs.hos…

FrnkNlsn's tweet image. Exact Fisher-Rao geodesics and guaranteed approximations of Fisher-Rao distances for multivariate normal distributions.

doi.org/10.1016/bs.hos…

Kundan Kumar reposted

The dynamics of a system of rods (multiple pendulums) is an ODE evolution on a 2nd-order algebraic variety. Defines highly complex patterns as soon as n>2. en.wikipedia.org/wiki/Double_pe…


Kundan Kumar reposted

A throwback to some algorithmic botany #rstats #creativecoding #plottertwitter

dickie_roper's tweet image. A throwback to some algorithmic botany
#rstats #creativecoding #plottertwitter

Kundan Kumar reposted

Oldies but goldies: Nina Amenta, Marshall Bern, Manolis Kamvysselis, A New Voronoi-Based Surface Reconstruction Algorithm, 1998. Introduces the power-crust algorithm, one of the first reconstruction methods with theoretical guarantees. web.cs.ucdavis.edu/~amenta/pubs/c…

gabrielpeyre's tweet image. Oldies but goldies: Nina Amenta, Marshall Bern, Manolis Kamvysselis, A New Voronoi-Based Surface Reconstruction Algorithm, 1998. Introduces the power-crust algorithm, one of the first reconstruction methods with theoretical guarantees. web.cs.ucdavis.edu/~amenta/pubs/c…

Kundan Kumar reposted

Every technical person knows about ordinary least-squares (OLS) but most don’t know *total* least-squares (TLS). These measure fitting error differently: OLS minimizes sum of sq. vertical distances whereas TLS minimizes the sum of orthogonal distances from data to fit line 1/2

docmilanfar's tweet image. Every technical person knows about ordinary least-squares (OLS) but most don’t know *total* least-squares (TLS).

These measure fitting error differently: OLS minimizes sum of sq. vertical distances whereas TLS minimizes the sum of orthogonal distances from data to fit line

1/2

Kundan Kumar reposted

The Kabsch-Nadas formula solves in closed form the orthogonal least square problem (aka orthogonal Procrustes). At the heart of the iterative closest point method for registration. en.wikipedia.org/wiki/Orthogona… en.wikipedia.org/wiki/Kabsch_al…


Loading...

Something went wrong.


Something went wrong.