Posts

Projects

*
AlgoTree: Comprehensive Tree Structure Algorithms in Python

AlgoTree: Comprehensive Tree Structure Algorithms in Python

Explore the AlgoTree package, a suite of utilities for working with tree-like data structures in Python using a generic API (duck typing).

ollama_data_tools: Tools for working with Ollama model data.

ollama_data_tools: Tools for working with Ollama model data.

Ollama Data Tools is a Python package and set of command line tools for working with Ollama model data, including searching, filtering, exporting,a and adapting for use with other inference engines like llamacpp.

Autoregressive Models: Inductive Biases and Projections

Autoregressive Models: Inductive Biases and Projections

This paper explores inductive biases and projection functions to enhance out-of-distribution generalization in autoregressive models. With the theory in place, we aim to implement these ideas in Python and develop tests and evaluations to validate our approach.

Bernoulli model

Bernoulli model

A general framework for constructing and working with Bernoulli models.

likelihood.model R package

likelihood.model R package

An API for constructing and working with likelihood models; works well with algebraic.mle R package

Rate-distorted perfect hash filter

A model of a second-order Bernoulli set with false positives and false negatives

algebraic.mle R package

algebraic.mle R package

An API for algebraic maximum likelihood fitting

femtograd R package

femtograd R package

An R package for backpropagation through a computational graph

Reliability Estimation in Series Systems

Reliability Estimation in Series Systems

Investigates maximum likelihood techniques to estimate component reliability from masked failure data in series systems. A simulation study is used to assess the sensitivity of the MLE to various parameters like sample size, masking probability, and other so on, using components with Weibull lifetimes arranged in series.

wei.series.md.c1.c2.c3

wei.series.md.c1.c2.c3

R package in narrow support of my master’s project in reliability estimation for series systems from masked failure data like the component cause of failure being unknown or the failure being censored.

algebraic.dist R package

algebraic.dist R package

An API for an algebra over distributions

R Package: Dynamic failure rate (DFR) distributions

R Package: Dynamic failure rate (DFR) distributions

An API for constructing and working with dynamic failure rates.

R package: md.tools

R package: md.tools

An API for constructing and working with likelihood models; works well with algebraic.mle R package

Homomorphic computational extensions

Homomorphic computational extensions

Class of homomorphic monads that extend algebraic data types in some well-defined way for faciliating faster or more accurate calculations, such as the logarithm trick.

Research projects

Problem sets

(and hopefully their solutions)

Alex Towell

Alex Towell

SIUe

Biography

Interests
  • AI / ML / LLMs
  • Cryptography and privacy

    • Encrypted search / homomorphic encryption / hash functions
    • Oblivious data structures and algorithms
  • Programming

    • Probabilistic data structures and algorithms
    • Programming language design / DSLs
  • Statistics / probability / data science / information theory
  • Information retrieval

    • Secure indexes
Education
  • M.S. Computer Science, 2015

    SIUe

  • M.S. Mathematics and Statistics, 2023

    SIUe

  • B.S. in Computer Science, 2011

    SIUe

Skills

R

90%

Mathematics

100%

Statistics

100%

Operations research

100%

Computer science

100%

Javascript

50%

Linux

90%

Python

70%

C/C++

90%

LaTeX

80%

Cloud computing

30%

Shell scripting

50%

Lisp

40%

Papers

On moral responsibility

On moral responsibility

Are we moral agents? Is reality compatible with the concept of moral agency?