papers ml-for-plse program-representation aspiring-academics minmax-optimization sigtbd fun cognitive-neuroscience constrained-decoding adversarial-nlp psycholinguistics concurrency-bug-detection thesis
-
Driving and suppressing the human language network using large language models
Generating sentences that control language system activations in the human brain
-
Understanding Computer Programs - Computational and Cognitive Perspectives
My PhD thesis
-
AutoAcademic
Get two LLMs to generate a paper that will be accepted at a conference of your choice
-
RaceInjector - Injecting Races To Evaluate And Learn Dynamic Race Detection Algorithms
Evaluating race detection algorithms proposed over the last four decades
-
GOLI - Goal-Optimized Linguistic Stimuli for Psycholinguistics and Cognitive Neuroscience
Automated generation of psycholinguistic stimuli
-
Convergent representations of computer programs in human and artificial neural networks
What aspects of computer programs are represented in the human brain during program comprehension?
-
CLAWSAT - Towards both robust and accurate code models
Can adversarial codes be used to improve the generalization and robustness of pre-trained code models? Yes.
-
Can cognitive neuroscience inform neuro-symbolic models?
Can recent studies which improve our understanding of how the human brain is involved in processing language possibly help inform neural-symbolic language models?
-
Generating adversarial programs
A differentiable generator of adversarial computer programs which can deceive ML models trained on computer programs
-
Comprehension of computer code relies primarily on domain-general executive brain regions
Which parts of our brains light up when reading programs? Is it the language region? Are programs treated as natural languages by the brain? Turns out, no.
-
Dependency-based neural representations for classifying lines of programs
How can ML models be used to represent lines of programs? We design a dependency-graph based neural representation of programs, which evaluates whether a given line of code has a vulnerability in it or not.
-
Human Baseline for Zero-Shot Transfer Learning of Good Dogs
How good are humans at zero-shot identification of dogs?
-
Aspiring academics
If you're an undergraduate/recent undergrad who wants to understand what doing research is all about, resources on this page should help you get a sense for it.
-
On the application of Danskin's to derivative-free minimax problems
We show how evolution strategies, which are stochastic gradient approximators, can be used to solve min-max problems.
-
Aspiring academics - older page
If you're an undergraduate/recent undergrad who wants to understand what doing research is all about, resources on this page should help you get a sense for it.
-
Data science for kids
At Aspiring Minds, we also started a fun side project where we introduced concepts in data science to high school students. It was a great deal of fun putting together fun exercises and a lesson plan. Check out our webpage datasciencekids.org for details!
-
Evaluating human skills using machine learning
At Aspiring Minds, we showed how the problem of quantifying human skills can be cast as a problem in machine learning. Our CACM paper provides details on this framework, while our KDD works show how we evaluate computer programs using ML.