Pittsburgh, PA, United States of America

Filipe J Cabrita Condessa

This inventor holds 22 USPTO granted patents and 22 published patent applications, primarily in Machine Learning. Top assignees: Robert Bosch, Carnegie Mellon University. Active years: 2022-2026.

IDiyas Innovation Intelligence. (2026). Inventor Profile: Filipe J Cabrita Condessa. Retrieved from https://idiyas.com/inventor/filipe-j-cabrita-condessa

Data source: USPTO (United States Patent and Trademark Office) public patent records. Weekly synchronization. How IDiyas builds this profile

Calculating rank

USPTO Granted Patents = 22 (Patent Applications = 22)

% Patents Active = 90.9

Average Co-Inventor Count = 4.2

ph-index = 1

Forward Citations = 6(Granted Patents)

Company Filing History:

Carnegie Mellon University

patents

where one patent can have more than one assignee

Years Active: 2022-2026

Loading Chart...

Areas of Expertise:

Sound Prediction

Cyber Threat Identification

Root Cause Analysis

Bayesian Optimization

Machine Learning Models

Text to Image Generation

Data Augmentation

Adversarial Examples

Image Processing

Multimodal Fusion

Active Learning

Community Detection

22 patents (USPTO):Sort by Forward CitationsExplore Patents

1. 12743620 - Predicting sound pleasantness using regression prediction machine learning model (2026-09-22),

2. 12587558 - System and method of artificial intelligence assisted cyber threat identification via webserver logs (2026-03-24),

3. 12536437 - Systems and methods for expert guided semi-supervision with contrastive loss for machine learning models (2026-01-27),

4. 12536300 - System and method for query efficient blackbox physically realizable attack with bayesian optimization (2026-01-27),

5. 12535803 - Methods and systems for performing root cause analysis with a trained machine learning model (2026-01-27),

6. 12464018 - System and method for high-resolution blackbox patch attack with Bayesian optimization (2025-11-04),

7. 12462159 - Methods and systems for training a machine learning model with measurement data captured during manufacturing process (2025-11-04),

8. 12387402 - System and method for localized text to image generation for free via cross attention (2025-08-12),

9. 12387057 - System and method for utilizing weak learners on large language models (2025-08-12),

10. 12277696 - Data augmentation for domain generalization (2025-04-14),

11. 12210619 - Method and system for breaking backdoored classifiers through adversarial examples (2025-01-28),

12. 12175336 - System and method for utilizing perturbation in a multimodal environment (2024-12-24),

13. 12172670 - Performance of neural networks under distribution shift (2024-12-24),

14. 12118788 - Learning semantic segmentation models in the absence of a portion of class labels (2024-10-15),

15. 12026621 - Method and system for low-query black-box universal attacks (2024-07-02),

Title: Innovations of Filipe J Cabrita Condessa

Introduction

Filipe J Cabrita Condessa is an accomplished inventor based in Pittsburgh, PA. He holds a remarkable portfolio of 15 patents, showcasing his expertise in advanced technologies and innovative solutions. His work primarily focuses on the intersection of language models and image generation.

Latest Patents

Among his latest patents is a "System and method for localized text to image generation for free via cross attention." This method involves receiving text data that describes both an object and a scene, utilizing a model with cross-attention layers to generate a prompt. The process includes generating a location mask and a cross-attention map to output a final image. Another significant patent is the "System and method for utilizing weak learners on large language models." This computer-implemented method converts tabular data into a text representation and utilizes a large language model to generate natural language descriptions and predictions based on queries.

Career Highlights

Filipe has worked with notable organizations such as Robert Bosch GmbH and Carnegie Mellon University. His experience in these institutions has contributed significantly to his development as an inventor and researcher in the field of technology.

Collaborations

Filipe has collaborated with talented individuals, including Jeremy Zico Kolter and Wan-Yi Lin. These partnerships have fostered innovation and creativity in his projects.

Conclusion

Filipe J Cabrita Condessa is a prominent figure in the realm of innovation, with a strong focus on language models and image generation. His contributions through patents and collaborations continue to influence the technological landscape.

Profile summary based on public USPTO records.

Data Sources: USPTO Patent Grant XML, Patent Center, EPO & CIPO • Normalized by IDiyas Innovation Graph. Methodology & provenance architecturePlease report any incorrect information to [email protected]

22 patents pending: