About

Antonio Ortega is the Dean’s Professor of Electrical and Computer Engineering at the University of Southern California (USC), where he has been a faculty member since 1994. He received his Telecommunications Engineering degree from Universidad Politecnica de Madrid in 1989 and earned his Ph.D. in Electrical Engineering from Columbia University in 1994, supported by a Fulbright scholarship. His research interests encompass signal compression, representation, communication, and analysis, with recent focus areas including distributed compression, multiview coding, error-tolerant compression, information representation in wireless sensor networks, and graph signal processing. Throughout his career at USC, Ortega has served as Associate Chair of EE-Systems and director of the Signal and Image Processing Institute. He is a Fellow of the IEEE, and a member of ACM and APSIPA. His professional service includes roles such as Chair of the Image and Multidimensional Signal Processing technical committee, member of the Board of Governors of the IEEE Signal Processing Society, and chair of the SPS Big Data Special Interest Group. He has held editorial positions with IEEE Transactions on Image Processing, IEEE Signal Processing Magazine, and is the inaugural Editor-in-Chief of the APSIPA Transactions on Signal and Information Processing. Ortega has received numerous awards, including the NSF CAREER award, IEEE Communications Society Leonard G. Abraham Prize Paper Award, and several best paper awards. His work has led to over 300 publications, several patents, and has been funded by agencies such as NSF, NASA, DOE, and companies including HP, Samsung, Google, and Chevron.

Research topics

• Computer Science
• Theoretical computer science
• Data science
• Physics
• Mathematics

Selected publications

• Digraph Signal Processing via Polar Decomposition

  2026-04-21

  articleSenior author

  We develop a signal processing framework for directed graphs. Unlike in the undirected setting, the shift operator of a directed graph generally lacks an orthogonal eigenbasis, making it difficult to define a Fourier transform with interpretable components. Leveraging the polar decomposition, we obtain two complementary eigendecompositions and introduce a graph Fourier transform that jointly encodes spectral responses in the corresponding eigenbases. This leads to a natural definition of convolution, which is lossless (recovering the original graph operator), and reduces to classical graph signal processing in the undirected case. Numerical experiments demonstrate applications of the framework to denoising.

  PublisherDOI

• Uncertainty Principle for Vertex-Time Graph Signal Processing

  Open MIND · 2026-02-03

  preprintSenior author

  We present an uncertainty principle for graph signals in the vertex-time domain, unifying the classical time-frequency and graph uncertainty principles within a single framework. By defining vertex-time and spectral-frequency spreads, we quantify signal localization across these domains. Our framework identifies a class of signals that achieve maximum concentration in both the spatial and temporal domains. These signals serve as fundamental atoms for a new vertex-time dictionary, enhancing signal reconstruction under practical constraints, such as intermittent data commonly encountered in sensor and social networks. Furthermore, we introduce a novel graph topology inference method leveraging the uncertainty principle. Numerical experiments on synthetic and real datasets validate the effectiveness of our approach, demonstrating improved reconstruction accuracy, greater robustness to noise, and enhanced graph learning performance compared to existing methods.

  DOI

• Uncertainty Principle for Vertex-Time Graph Signal Processing

  ArXiv.org · 2026-02-03

  articleOpen accessSenior author

  We present an uncertainty principle for graph signals in the vertex-time domain, unifying the classical time-frequency and graph uncertainty principles within a single framework. By defining vertex-time and spectral-frequency spreads, we quantify signal localization across these domains. Our framework identifies a class of signals that achieve maximum concentration in both the spatial and temporal domains. These signals serve as fundamental atoms for a new vertex-time dictionary, enhancing signal reconstruction under practical constraints, such as intermittent data commonly encountered in sensor and social networks. Furthermore, we introduce a novel graph topology inference method leveraging the uncertainty principle. Numerical experiments on synthetic and real datasets validate the effectiveness of our approach, demonstrating improved reconstruction accuracy, greater robustness to noise, and enhanced graph learning performance compared to existing methods.

  PublisherOA PDF

• AutoML for multi-class anomaly compensation of sensor drift

  Measurement · 2025-03-03 · 7 citations

  articleOpen access

  Addressing sensor drift is essential in industrial measurement systems, where precise data output is necessary for maintaining accuracy and reliability in monitoring processes, as it progressively degrades the performance of machine learning models over time. Our findings indicate that the standard cross-validation method used in existing model training overestimates performance by inadequately accounting for drift. This is primarily because typical cross-validation techniques allow data instances to appear in both training and testing sets, thereby distorting the accuracy of the predictive evaluation. As a result, these models are unable to precisely predict future drift effects, compromising their ability to generalize and adapt to evolving data conditions. This paper presents two solutions: (1) a novel sensor drift compensation learning paradigm for validating models, and (2) automated machine learning (AutoML) techniques to enhance classification performance and compensate sensor drift. By employing strategies such as data balancing, meta-learning, automated ensemble learning, hyperparameter optimization, feature selection, and boosting, our AutoML-DC (Drift Compensation) model significantly improves classification performance against sensor drift. AutoML-DC further adapts effectively to varying drift severities. • Our analysis demonstrates that the conventional training configurations are suboptimal in learning and compensating for sensor drift. Thus, we propose a novel sensor drift compensation learning training paradigm that closely matches real-world scenarios. • Our findings further indicate, that AutoML techniques along with the proposed training paradigm enable effective drift adaptation to evolving levels of drift severity and complex drift dynamics in sensor measurements. • By utilizing meta-learning, AutoML-DC starts from initial configurations based on prior data, lowering the extensive data requirements normally needed for training neural networks or ensemble models. • We make use of AutoML techniques to enhance model robustness (see standard deviation) by combining multiple models to capture diverse data patterns, optimizing feature selection, and preventing overfitting through smart training termination. • We conduct extensive benchmarking experiments against existing models and highlight the significant accuracy improvements realized when adopting AutoML-DC in practical drift compensation scenarios in industrial measurements.

  PublisherDOI

• Ciática y lumbalgia. Dos entidades distintas, ¿sabemos diferenciar ambos cuadros?

  2025-03-25

  article
  PublisherDOI

• Robust Classification under Noisy Labels: A Geometry-Aware Reliability Framework for Foundation Models

  ArXiv.org · 2025-07-31

  preprintOpen accessSenior author

  Foundation models (FMs) pretrained on large datasets have become fundamental for various downstream machine learning tasks, in particular in scenarios where obtaining perfectly labeled data is prohibitively expensive. In this paper, we assume an FM has to be fine-tuned with noisy data and present a two-stage framework to ensure robust classification in the presence of label noise without model retraining. Recent work has shown that simple k-nearest neighbor (kNN) approaches using an embedding derived from an FM can achieve good performance even in the presence of severe label noise. Our work is motivated by the fact that these methods make use of local geometry. In this paper, following a similar two-stage procedure, reliability estimation followed by reliability-weighted inference, we show that improved performance can be achieved by introducing geometry information. For a given instance, our proposed inference uses a local neighborhood of training data, obtained using the non-negative kernel (NNK) neighborhood construction. We propose several methods for reliability estimation that can rely less on distance and local neighborhood as the label noise increases. Our evaluation on CIFAR-10 and DermaMNIST shows that our methods improve robustness across various noise conditions, surpassing standard K-NN approaches and recent adaptive-neighborhood baselines.

  PublisherOA PDFDOI

• Joint Optimization of Primary and Secondary Transforms Using Rate-Distortion Optimized Transform Design

  ArXiv.org · 2025-05-21

  preprintOpen access

  Data-dependent transforms are increasingly being incorporated into next-generation video coding systems such as AVM, a codec under development by the Alliance for Open Media (AOM), and VVC. To circumvent the computational complexities associated with implementing non-separable data-dependent transforms, combinations of separable primary transforms and non-separable secondary transforms have been studied and integrated into video coding standards. These codecs often utilize rate-distortion optimized transforms (RDOT) to ensure that the new transforms complement existing transforms like the DCT and the ADST. In this work, we propose an optimization framework for jointly designing primary and secondary transforms from data through a rate-distortion optimized clustering. Primary transforms are assumed to follow a path-graph model, while secondary transforms are non-separable. We empirically evaluate our proposed approach using AVM residual data and demonstrate that 1) the joint clustering method achieves lower total RD cost in the RDOT design framework, and 2) jointly optimized separable path-graph transforms (SPGT) provide better coding efficiency compared to separable KLTs obtained from the same data.

  PublisherOA PDFDOI

• Sparse Interpretable Deep Learning with LIES Networks for Symbolic Regression

  ArXiv.org · 2025-06-09

  preprintOpen access

  Symbolic regression (SR) aims to discover closed-form mathematical expressions that accurately describe data, offering interpretability and analytical insight beyond standard black-box models. Existing SR methods often rely on population-based search or autoregressive modeling, which struggle with scalability and symbolic consistency. We introduce LIES (Logarithm, Identity, Exponential, Sine), a fixed neural network architecture with interpretable primitive activations that are optimized to model symbolic expressions. We develop a framework to extract compact formulae from LIES networks by training with an appropriate oversampling strategy and a tailored loss function to promote sparsity and to prevent gradient instability. After training, it applies additional pruning strategies to further simplify the learned expressions into compact formulae. Our experiments on SR benchmarks show that the LIES framework consistently produces sparse and accurate symbolic formulae outperforming all baselines. We also demonstrate the importance of each design component through ablation studies.

  PublisherOA PDFDOI

• MEGA (Megaplicatura Endoscópica Gástrica): nueva variante del POSE 2.0 con doble G-LIX para el tratamiento endoscópico de la obesidad

  2025-07-24

  article
  PublisherDOI

• Select Azo Compounds Post-translationally Modulate HTRA1 Abundance and Activity Potentially through Interactions at the Trimer Interface

  ACS Chemical Biology · 2025-07-30

  article

  High-temperature requirement protein A1 (HTRA1) is a secreted serine protease with diverse substrates, including extracellular matrix proteins, proteins involved in amyloid deposition, and growth factors. Accordingly, HTRA1 has been implicated in a variety of neurodegenerative diseases including a leading cause of blindness in the elderly, age-related macular degeneration (AMD). In fact, genomewide association studies have identified that the 10q26 locus that contains HTRA1 confers the strongest genetic risk factor for AMD. A recent study has suggested that AMD-associated risk alleles located in the HTRA1 gene correlate with a significant age-related defect in HTRA1 synthesis in the retinal pigmented epithelium (RPE) within the eye, possibly accounting for AMD susceptibility. Thus, we sought to identify small molecule enhancers of HTRA1 transcription and/or protein abundance using an unbiased high-throughput screening approach. To accomplish this goal, we used CRISPR/Sp.Cas9 engineering to introduce an 11-amino-acid luminescent peptide tag (HiBiT) onto the C-terminus of HTRA1 in immortalized ARPE-19 cells. Editing was very efficient (∼88%), verified by genomic DNA analysis, short interfering RNA (siRNA), and HiBiT blotting. A total of 1920 compounds from two libraries were screened. An azo compound with reported antiamyloidogenic and cardioprotective activity, Chicago Sky Blue 6B (CSB), was identified as an enhancer of endogenous HTRA1 secretion (2.0 ± 0.3 fold) and intracellular levels (1.7 ± 0.2 fold). These results were counter-screened using HiBiT complement factor H (CFH) edited ARPE-19 cells, verified using HiBiT blotting, and were not due to HTRA1 transcriptional changes. Importantly, serine hydrolase activity-based protein profiling (SH-ABPP) demonstrated that CSB does not affect HTRA1’s specific activity. However, interestingly, in follow-up studies, Congo Red, another azo compound structurally similar to CSB, also substantially increased intracellular HTRA1 levels (up to 3.6 ± 0.3 fold) but was found to significantly impair HTRA1 enzymatic reactivity (0.45 ± 0.07 fold). Computational modeling of potential azo dye interaction with HTRA1 suggests that CSB and Congo Red can bind to the noncatalytic face of the trimer interface but with different orientation tolerances and interaction energies. These studies identify select azo dyes as HTRA1 chemical probes that may serve as starting points for future HTRA1-centered small molecule therapeutics.

  PublisherDOI

Recent grants

• CIF: Small: Graph Signal Processing Methods for Data-driven System Design

  NSF · $500k · 2020–2024

• CIF: Small: Graph Signal Sampling: Theory and Applications

  NSF · $499k · 2015–2019

• CIF: Small: Wavelets on Graphs - Theory and Applications

  NSF · $500k · 2010–2014

Frequent coauthors

• Ali H. Sayed

  École Polytechnique Fédérale de Lausanne

  5012 shared

• Athina P. Petropulu

  Rutgers, The State University of New Jersey

  4816 shared

• Sergios Theodoridis

  National and Kapodistrian University of Athens

  4816 shared

• Paris Smaragdis

  4814 shared

• Shoji Makino

  Waseda University

  4814 shared

• Béatrice Pesquet‐Popescu

  University of Maryland, Baltimore County

  4758 shared

• Ahmed H. Tewfik

  Apple (United Kingdom)

  4228 shared

• Fernando Pereira

  4096 shared

Education

• PhD, Electrical Engineering

  Columbia University

  1994

Awards & honors

• Fellow of the IEEE
• NSF CAREER award
• 1997 IEEE Communications Society Leonard G. Abraham Prize Pa…
• IEEE Signal Processing Society 1999 Magazine Award
• 2006 EURASIP Journal of Advances in Signal Processing Best P…