Author Archives: snikolov

Generative rhythm as a self-avoiding random walk

A rhythm is just a binary vector of a certain size with some of the bits turned on. The layering of rhythms is the generation of new binary vectors (each representing a new percussive instrument) that “go well” with the

Generative rhythm as a self-avoiding random walk

A rhythm is just a binary vector of a certain size with some of the bits turned on. The layering of rhythms is the generation of new binary vectors (each representing a new percussive instrument) that “go well” with the

Early detection of Twitter trends explained

A couple of weeks ago on Halloween night, I was out with some friends when my advisor sent me a message to check web.mit.edu, right now. It took me a few seconds of staring to realize that an article about

Early detection of Twitter trends explained

A couple of weeks ago on Halloween night, I was out with some friends when my advisor sent me a message to check web.mit.edu, right now. It took me a few seconds of staring to realize that an article about

Information Diffusion on Twitter

This spring, I volunteered to teach a lecture in a new Berkeley course called “Analyzing Big Data With Twitter,” developed jointly by Twitter and Berkeley’s School of Information. I had recently done my masters thesis work on predicting the spread

Information Diffusion on Twitter

This spring, I volunteered to teach a lecture in a new Berkeley course called “Analyzing Big Data With Twitter,” developed jointly by Twitter and Berkeley’s School of Information. I had recently done my masters thesis work on predicting the spread

Semi-Supervised Shape Classification with Manifold Regularization

For my Statistical Learning Theory class I did a project on shape classification using manifold regularization. You can read the abstract below. You can also find the paper here and the code here. We approach the problem of semi-supervised shape classification

Semi-Supervised Shape Classification with Manifold Regularization

For my Statistical Learning Theory class I did a project on shape classification using manifold regularization. You can read the abstract below. You can also find the paper here and the code here. We approach the problem of semi-supervised shape classification

Underactuated Control of Vehicular Traffic

When self-driving robotic cars begin to share the road with regular cars, could we control the robotic cars to smooth out traffic jams? I did a numerical and theoretical study of vehicle traffic dynamics and control policies that smooth traffic even

Underactuated Control of Vehicular Traffic

When self-driving robotic cars begin to share the road with regular cars, could we control the robotic cars to smooth out traffic jams? I did a numerical and theoretical study of vehicle traffic dynamics and control policies that smooth traffic even

Stan Explains Things: RANdom SAmple Consensus

I’ve decided to start writing briefly and informally about technical things, so that I could understand them better, and so that someone else might get some insight.  I’ll be writing these under the appropriately vague title of “Stan Explains Things”.

Stan Explains Things: RANdom SAmple Consensus

I’ve decided to start writing briefly and informally about technical things, so that I could understand them better, and so that someone else might get some insight.  I’ll be writing these under the appropriately vague title of “Stan Explains Things”.

The Statistical Structure of Rhythm

I took a quick break with some random hacking last night. I wanted to see what kind of statistical structure there is in rhythm, since I am always tapping on various vaguely percussive objects. You can see the code on

The Statistical Structure of Rhythm

I took a quick break with some random hacking last night. I wanted to see what kind of statistical structure there is in rhythm, since I am always tapping on various vaguely percussive objects. You can see the code on

You Shouldn’t Be Working

If you’re an ambitious person you’re probably very familiar with that voice in the back of your head that says you should be working right now. There are so many fun things you could be doing. There’s a party tonight.

You Shouldn’t Be Working

If you’re an ambitious person you’re probably very familiar with that voice in the back of your head that says you should be working right now. There are so many fun things you could be doing. There’s a party tonight.

Do Something That Moves You

Before I die, I want to use my skills and talents to help make people’s lives better. I am privileged to have a lot of opportunities to do so.  So what do I do? A friend of mine who is

Do Something That Moves You

Before I die, I want to use my skills and talents to help make people’s lives better. I am privileged to have a lot of opportunities to do so.  So what do I do? A friend of mine who is

Metacreativity

A lot of people might say that automating a given creative task kills creativity and devalues the result. Why learn to sing if you can just use autotune? Why learn to draw portraits when you can apply a bunch of

Metacreativity

A lot of people might say that automating a given creative task kills creativity and devalues the result. Why learn to sing if you can just use autotune? Why learn to draw portraits when you can apply a bunch of

Finding Features in Fly Embryos

In progress … (but feel free to read anyway!) In this post I’ll tell you about some progress I’ve made in automatically detecting morphological features in fly embryo images.  But first, a little background. Why would one want to automatically

Finding Features in Fly Embryos

In progress … (but feel free to read anyway!) In this post I’ll tell you about some progress I’ve made in automatically detecting morphological features in fly embryo images.  But first, a little background. Why would one want to automatically

The Wrong Kind of Modesty

Many smart people tend to be modest about their abilities.  The most popular form of this modesty, in my experience is to say or think “I’m really not that smart, but I work hard.” Hard work is something people take

The Wrong Kind of Modesty

Many smart people tend to be modest about their abilities.  The most popular form of this modesty, in my experience is to say or think “I’m really not that smart, but I work hard.” Hard work is something people take

Robotic Gymnastics

It’s that final project time of year again at MIT, and I thought I’d write a few posts about the projects I’ll be working on for the next month and a half.  This first one is for my class 6.832

Robotic Gymnastics

It’s that final project time of year again at MIT, and I thought I’d write a few posts about the projects I’ll be working on for the next month and a half.  This first one is for my class 6.832

Stuck detection for mobile robots

No matter how clever your navigation and obstacle avoidance code is, occasionally, your robot will probably get stuck.  Perhaps it bashes into a wall it didn’t see when it goes for a target, or maybe some appendage got snagged somewhere. 

Stuck detection for mobile robots

No matter how clever your navigation and obstacle avoidance code is, occasionally, your robot will probably get stuck.  Perhaps it bashes into a wall it didn’t see when it goes for a target, or maybe some appendage got snagged somewhere. 

Wall-following for mobile robots

Wall-following is a relatively simple and useful method for an autonomous mobile robot to explore its environment. However, wall-following can be tricky for a number of reasons: Incomplete sensor coverage Limiting behaviors of range-finding infrared (IR) sensors Variety and complexity

Wall-following for mobile robots

Wall-following is a relatively simple and useful method for an autonomous mobile robot to explore its environment. However, wall-following can be tricky for a number of reasons: Incomplete sensor coverage Limiting behaviors of range-finding infrared (IR) sensors Variety and complexity

Teaching People to Teach Machines with Mechanical Turk

Mechanical Turk has proven to be a powerful tool for machine learning.  In particular, it makes it very easy to generate large amounts of training data for machine learning tasks.  For example, one can have Mechanical Turk workers transcribe recorded

Teaching People to Teach Machines with Mechanical Turk

Mechanical Turk has proven to be a powerful tool for machine learning.  In particular, it makes it very easy to generate large amounts of training data for machine learning tasks.  For example, one can have Mechanical Turk workers transcribe recorded

Principal Component Analysis and Extensions

Last summer, while working at Numenta, I spent some time reading about sensory coding and how it relates to pattern recognition.  I read a lot about sparse coding, principal component analysis, independent component analysis, nonnegative matrix factorization and other methods.  It

Principal Component Analysis and Extensions

Last summer, while working at Numenta, I spent some time reading about sensory coding and how it relates to pattern recognition.  I read a lot about sparse coding, principal component analysis, independent component analysis, nonnegative matrix factorization and other methods.  It

On the perceptual accessibility of abstract physical laws

There’s a quote widely attributed to Feynman that goes like this: “If you think you understand quantum mechanics, you don’t understand quantum mechanics.” Yet quantum mechanics, as weird as it is, is considered to be the most experimentally accurate theory

On the perceptual accessibility of abstract physical laws

There’s a quote widely attributed to Feynman that goes like this: “If you think you understand quantum mechanics, you don’t understand quantum mechanics.” Yet quantum mechanics, as weird as it is, is considered to be the most experimentally accurate theory

The Principal Components of Handwritten ‘2’s

Another example of principal components of shape datasets, this time images of ‘2’s from the MNIST handwritten digit database.  The images below are the first 50 principal components.

The Principal Components of Handwritten ‘2’s

Another example of principal components of shape datasets, this time images of ‘2’s from the MNIST handwritten digit database.  The images below are the first 50 principal components.

The Principal Components of Butterflies

The first 50 principal components of a dataset of signed distance functions computed from 100 binary images (“silhouettes”) of butterflies.

The Principal Components of Butterflies

The first 50 principal components of a dataset of signed distance functions computed from 100 binary images (“silhouettes”) of butterflies.