>_ Tibor Sloboda

Research

My research in multimodal systems and generative models, earlier work in medicine and chemistry, and selected publications.

I am a PhD candidate in artificial intelligence at FIIT, Slovak University of Technology, working on multimodal systems and generative models. I am interested in how models represent and combine different kinds of information, and in the mathematics of those relationships. My work on cross-modal compatibility examines that mathematical foundation.

Earlier work in medicine & chemistry

Much of my earlier research focused on biomedical ML. In histology, I worked on transforming stain images with unpaired GANs, including editable transformations and attention-enhanced models. I also contributed to PCD quant, which uses automatic image analysis to assess ciliary ultrastructure.

In chemistry, I worked on machine learning for phototoxicity prediction and investigated the descriptors used as model inputs. I also helped build and maintain MLTox, the online application that made those methods available to users.

Research & engineering

I have also co-authored work on guarded query routing for LLMs. Alongside research, I work on model optimization and AI infrastructure for heterogeneous hardware. A method's usefulness depends on the data it will encounter, the compute it needs, and the conditions under which it will run.

From a model score to a decision

Prediction tasks such as phototoxicity assessment raise a practical question: how should a model's output become a decision? For a binary classifier, a threshold determines which scores count as positive. In the example below, lowering it catches more positives and produces more false alarms.

The synthetic example below makes that tradeoff visible. Drag the red threshold τ to the left, then to the right, or focus it and use the arrow keys. Watch recall and the false positive rate change together, and the operating point move along the same ROC curve.

Decision threshold explorer: score distributions and ROC curve Left: two Gaussian classifier score distributions - negatives at mean 0, positives at mean 1.8, both unit variance - with a movable decision threshold at tau = 0.9. Scores above the threshold are called positive; the shaded tails are the true positives and false positives that choice creates. Right: the ROC curve those same two distributions trace as the threshold sweeps, AUC = 0.898, with the current operating point marked. Dragging the threshold moves the operating point and recomputes recall, false positive rate, precision, and F1 from the normal CDF. -2 0 2 4 classifier score s negatives · N(0, 1) positives · N(1.8, 1) τ = 0.90 0 0 0.5 0.5 1 1 false positive rate true positive rate AUC = 0.898

recall 0.816fpr 0.184precision 0.655F1 0.727prevalence 30%d′ = 1.8

fig. 1 · synthetic Gaussian score distributions with unit variance and separation d′ = 1.8. Left: scores above τ count as positive; teal marks true positives and red marks false positives. Right: the corresponding ROC curve (analytic AUC = 0.898) and current operating point. Recall = Φ(d′−τ), false positive rate = 1−Φ(τ), and precision assumes 30% prevalence. Without JavaScript, τ = 0.90.

Moving the threshold changes the decisions and their errors, even though the model and its ROC AUC stay the same. Choosing a threshold for an application requires evaluation on relevant data, calibration checks, and an understanding of what each kind of error costs. These distributions are synthetic; they are not results from a deployed medical system.

Publications & preprints

Google Scholar has my publication record and citation information.

For a research collaboration or help evaluating a model, get in touch.

Azu

AI agent

1 of 3 · Introduction

Chat about this page

Azu is an AI agent that runs locally on your device. You can chat with Azu and ask questions about the current page. No chat data leaves your device.