A5033615917- Machine Learning and Algorithms
- Medical Coding and Health Information
- Software Testing and Debugging Techniques
- Imbalanced Data Classification Techniques
- Adversarial Robustness in Machine Learning
- Internet Traffic Analysis and Secure E-voting
- AI in cancer detection
- Neural Networks and Applications
- Natural Language Processing Techniques
- Cryptography and Data Security
- Machine Learning and Data Classification
- Ferroelectric and Negative Capacitance Devices
- Formal Methods in Verification
- Privacy-Preserving Technologies in Data
- semigroups and automata theory
- Reliability and Agreement in Measurement
Universidad ORT Uruguay
2018-2022
Abstract Background In classification and diagnostic testing, the receiver-operator characteristic (ROC) plot area under ROC curve (AUC) describe how an adjustable threshold causes changes in two types of error: false positives negatives. Only part AUC are informative however when they used with imbalanced data. Hence, alternatives to have been proposed, such as partial precision-recall curve. However, these cannot be fully interpreted AUC, because ignore some information about actual...
This paper presents a novel on-the-fly, black-box, property-checking through learning approach as means for verifying requirements of recurrent neural networks (RNN) in the context sequence classification. Our technique steps on tool probably approximately correct (PAC) deterministic finite automata (DFA). The classifier inside black-box consists Boolean combination several components, including RNN under analysis together with to be checked, possibly modeled themselves. On one hand, if...
We define a congruence that copes with null next-symbol probabilities arise when the output of language model is constrained by some means during text generation. develop an algorithm for efficiently learning quotient respect to this and evaluate it on case studies analyzing statistical properties LLM.
This work studies the question of learning probabilistic deterministic automata from language models. For this purpose, it focuses on analyzing relations defined algebraic structures over strings by equivalences and similarities probability distributions. We introduce a congruence that extends classical Myhill-Nerode for formal languages. new is basis defining regularity present an active algorithm computes quotient with respect to whenever model regular. The paper also defines notion...
This paper explores the use of Private Aggregation Teacher Ensembles (PATE) in a setting where students have their own private data that cannot be revealed as is to ensemble. We propose privacy model introduces local differentially mechanism protect student data. implemented and analyzed it case studies from security health domains, result experiment was twofold. First, this does not significantly affecs predictive capabilities, second, unveiled interesting issues with so-called dependency...
We propose a new active learning algorithm for PDFA based on three main aspects: congruence over states which takes into account next-symbol probability distributions, quantization that copes with differences in and an efficient tree-based data structure. Experiments showed significant performance gains respect to reference implementations.