Tim Johnsen Email & Phone Number

Irvine, Ca, Us

Irvine, Ca, Us

Irvine, Ca, Us

My working thesis is entitled “Adaptable Dynamic Deep Neural Networks”.Executing state-of-the-art (SoA) neural network models can have large resource costs in the form of execution time and energy expenditure, due to an increasingly large amount of parameters (e.g., activation functions, bias terms, and weights) that results in upwards to billions of FLOPS. An undesirable, but inherent, feature of such neural network models is that they have a static number of parameters which must match that required to solve the most difficult, expected scenario. This creates a problem that when the model encounters a less difficult scenario, the model is executed using a sub-optimal amount of resources. I am researching a solution that uses a dynamic neural architecture that adapts the number of active model parameters to the perceived context of a given scenario.

Berkeley, Ca, Us

I am developing machine learning applications for uncertainty quantification in measurements of watershed environments, for the purpose of improved decision making from a Self Guided Field Lab (SGFL) designed to study evapotranspiration systems. My current contribution is a Denoisoing AutoEncoder with Monte Carlo sampling methods, that repairs corrupted sensor observations to predict evapotranspiration (ET) from a downstream model. Further, I use spatial-temporal multiband sensors from various sources to both predict ET and help decide sensor placement in the development phase of observation towers, satellites, and other devices.

San Diego, Ca, Us

First year of studies.

Santa Clara, California, Us

My main focus was on recommendation systems for healthcare plans. I researched using Graph Convolutional Neural Networks (GCNN) with customer and agent co-sale matrices, and semi supervised learning. My tasks varied from model development and comparison, data processing and pipeline development, to researching and analyzing new methods. I wrote a custom library in python for recommendations, to track various trained models and be able to rewind and reuse those models. This was a series of 2 remote, summer internships during the pandemic.

Mountain View, California, Us

My main focus is on applied machine learning and spectroscopy (with multimodal sensors), to identify, classify, and characterize minerals, rocks, organics, and exoplanets. This is a contractor position at NASA Ames Research Center, through SETI Institute. I have worked here through several NASA funded internships and other grants, both part time and full time, while completing my degrees. This resulted in several oral presentations, poster presentations, and conference abstracts, along with a journal publication as lead author and my undergraduate senior thesis.

Moffett Field, California, Us

Contract position with SETI institute.

Moffett Field, California, Us

My responsibilities were to help deploy an autonomous 2D-traverse wind tunnel probe, and develop a Java web-crawler for post processing ADS-B aircraft flyover information (from the ground up). This was a summer internship at the Fluid Mechanics Lab with Bruce Storms.

Santa Clara, California, Us

First summer internship dates.

Doctor Of Philosophy - Phd, Computational Sciences

Doctor Of Philosophy - Phd, Computational Science

Master Of Science - Ms, Data Analytics From The Department Of Applied Data Science

Bachelor Of Science - Bs, Engineering Physics/Applied Physics

Transfer, Computer Science