Search Results for author: Matias Valdenegro-Toro

Found 49 papers, 11 papers with code

Overconfidence is Key: Verbalized Uncertainty Evaluation in Large Language and Vision-Language Models

no code implementations5 May 2024 Tobias Groot, Matias Valdenegro-Toro

Language and Vision-Language Models (LLMs/VLMs) have revolutionized the field of AI by their ability to generate human-like text and understand images, but ensuring their reliability is crucial.

Object Counting

A Neuromorphic Approach to Obstacle Avoidance in Robot Manipulation

no code implementations8 Apr 2024 Ahmed Faisal Abdelrahman, Matias Valdenegro-Toro, Maren Bennewitz, Paul G. Plöger

To investigate the utility of brain-inspired sensing and data processing, we developed a neuromorphic approach to obstacle avoidance on a camera-equipped manipulator.

Event-based vision Robot Manipulation

Sanity Checks for Explanation Uncertainty

no code implementations25 Mar 2024 Matias Valdenegro-Toro, Mihir Mulye

Explanations for machine learning models can be hard to interpret or be wrong.

Uncertainty Quantification for Gradient-based Explanations in Neural Networks

no code implementations25 Mar 2024 Mihir Mulye, Matias Valdenegro-Toro

By computing the coefficient of variation of these distributions, we evaluate the confidence in the explanation and determine that the explanations generated using Guided Backpropagation have low uncertainty associated with them.

Uncertainty Quantification

Uncertainty Quantification for cross-subject Motor Imagery classification

1 code implementation14 Mar 2024 Prithviraj Manivannan, Ivo Pascal de Jong, Matias Valdenegro-Toro, Andreea Ioana Sburlea

We applied a variety of Uncertainty Quantification methods to predict misclassifications for a Motor Imagery Brain Computer Interface.

Classification Motor Imagery +1

ChatGPT Prompting Cannot Estimate Predictive Uncertainty in High-Resource Languages

no code implementations10 Nov 2023 Martino Pelucchi, Matias Valdenegro-Toro

This paper aims to join the growing literature regarding ChatGPT's abilities by focusing on its performance in high-resource languages and on its capacity to predict its answers' accuracy by giving a confidence level.

Sanity Checks for Saliency Methods Explaining Object Detectors

no code implementations4 Jun 2023 Deepan Chakravarthi Padmanabhan, Paul G. Plöger, Octavio Arriaga, Matias Valdenegro-Toro

Adebayo et al.'s work on evaluating saliency methods for classification models illustrate certain explanation methods fail the model and data randomization tests.

Object object-detection +1

DExT: Detector Explanation Toolkit

1 code implementation21 Dec 2022 Deepan Chakravarthi Padmanabhan, Paul G. Plöger, Octavio Arriaga, Matias Valdenegro-Toro

State-of-the-art object detectors are treated as black boxes due to their highly non-linear internal computations.


Disentangled Uncertainty and Out of Distribution Detection in Medical Generative Models

no code implementations11 Nov 2022 Kumud Lakara, Matias Valdenegro-Toro

Trusting the predictions of deep learning models in safety critical settings such as the medical domain is still not a viable option.

Image-to-Image Translation Out-of-Distribution Detection +2

Machine Learning Students Overfit to Overfitting

no code implementations7 Sep 2022 Matias Valdenegro-Toro, Matthia Sabatelli

Overfitting and generalization is an important concept in Machine Learning as only models that generalize are interesting for general applications.


Self-supervised Learning for Sonar Image Classification

1 code implementation20 Apr 2022 Alan Preciado-Grijalva, Bilal Wehbe, Miguel Bande Firvida, Matias Valdenegro-Toro

Self-supervised learning has proved to be a powerful approach to learn image representations without the need of large labeled datasets.

Classification Denoising +3

A Deeper Look into Aleatoric and Epistemic Uncertainty Disentanglement

no code implementations20 Apr 2022 Matias Valdenegro-Toro, Daniel Saromo

Uncertainty quantification is required for many applications, and disentangled aleatoric and epistemic uncertainties are best.

Disentanglement Uncertainty Quantification

Feature Disentanglement of Robot Trajectories

no code implementations6 Dec 2021 Matias Valdenegro-Toro, Daniel Harnack, Hendrik Wöhrle

Modeling trajectories generated by robot joints is complex and required for high level activities like trajectory generation, clustering, and classification.

Clustering Disentanglement

Benchmark for Out-of-Distribution Detection in Deep Reinforcement Learning

no code implementations5 Dec 2021 Aaqib Parvez Mohammed, Matias Valdenegro-Toro

Out of distribution detection for RL is generally not well covered in the literature, and there is a lack of benchmarks for this task.

Out-of-Distribution Detection Out of Distribution (OOD) Detection +2

Exploring the Limits of Epistemic Uncertainty Quantification in Low-Shot Settings

1 code implementation NeurIPS Workshop LatinX_in_AI 2021 Matias Valdenegro-Toro

Uncertainty quantification in neural network promises to increase safety of AI systems, but it is not clear how performance might vary with the training set size.

Out-of-Distribution Detection Uncertainty Quantification

The VVAD-LRS3 Dataset for Visual Voice Activity Detection

no code implementations28 Sep 2021 Adrian Lubitz, Matias Valdenegro-Toro, Frank Kirchner

With a Convolutional Neural Network Long Short Term Memory (CNN LSTM) on facial images an accuracy of 92% was reached on the test set.

Action Detection Activity Detection +2

Teaching Uncertainty Quantification in Machine Learning through Use Cases

no code implementations19 Aug 2021 Matias Valdenegro-Toro

Uncertainty in machine learning is not generally taught as general knowledge in Machine Learning course curricula.

BIG-bench Machine Learning General Knowledge +2

Forward-Looking Sonar Patch Matching: Modern CNNs, Ensembling, and Uncertainty

no code implementations2 Aug 2021 Arka Mallick, Paul Plöger, Matias Valdenegro-Toro

Application of underwater robots are on the rise, most of them are dependent on sonar for underwater vision, but the lack of strong perception capabilities limits them in this task.

Change Detection Patch Matching

Pre-trained Models for Sonar Images

1 code implementation2 Aug 2021 Matias Valdenegro-Toro, Alan Preciado-Grijalva, Bilal Wehbe

Machine learning and neural networks are now ubiquitous in sonar perception, but it lags behind the computer vision field due to the lack of data and pre-trained models specifically for sonar images.

Transfer Learning

I Find Your Lack of Uncertainty in Computer Vision Disturbing

no code implementations16 Apr 2021 Matias Valdenegro-Toro

Neural networks are used for many real world applications, but often they have problems estimating their own confidence.

Out-of-Distribution Detection Uncertainty Quantification

Unsupervised Difficulty Estimation with Action Scores

no code implementations23 Nov 2020 Octavio Arriaga, Matias Valdenegro-Toro

Evaluating difficulty and biases in machine learning models has become of extreme importance as current models are now being applied in real-world situations.

Image Classification object-detection +1

Black-Box Optimization of Object Detector Scales

no code implementations29 Oct 2020 Mohandass Muthuraja, Octavio Arriaga, Paul Plöger, Frank Kirchner, Matias Valdenegro-Toro

In this work, we propose the use of Black-box optimization methods to tune the prior/default box scales in Faster R-CNN and SSD, using Bayesian Optimization, SMAC, and CMA-ES.

Bayesian Optimization Object +1

Know Where To Drop Your Weights: Towards Faster Uncertainty Estimation

no code implementations NeurIPS Workshop ICBINB 2020 Akshatha Kamath, Dwaraknath Gnaneshwar, Matias Valdenegro-Toro

Through our experiments, we show a significant reduction in the GFLOPS required to model uncertainty, compared to Monte Carlo DropConnect, with marginal trade-off in performance.

Can Reinforcement Learning for Continuous Control Generalize Across Physics Engines?

no code implementations27 Oct 2020 Aaqib Parvez Mohammed, Matias Valdenegro-Toro

There are multiple algorithms that solve the task in a physics engine based environment but there is no work done so far to understand if the RL algorithms can generalize across physics engines.

Continuous Control reinforcement-learning +1

Hey Human, If your Facial Emotions are Uncertain, You Should Use Bayesian Neural Networks!

no code implementations17 Aug 2020 Maryam Matin, Matias Valdenegro-Toro

In this paper we show that Bayesian Neural Networks, as approximated using MC-Dropout, MC-DropConnect, or an Ensemble, are able to model the aleatoric uncertainty in facial emotion recognition, and produce output probabilities that are closer to what a human expects.

Facial Emotion Recognition

Data augmentation with Symbolic-to-Real Image Translation GANs for Traffic Sign Recognition

no code implementations17 Jul 2019 Nour Soufi, Matias Valdenegro-Toro

SqueezeNet is a good candidate for efficient image classification of traffic signs, but in our experiments it does not reach high accuracy, and we believe this is due to lack of data, requiring data augmentation.

Autonomous Driving Data Augmentation +3

Learning Objectness from Sonar Images for Class-Independent Object Detection

no code implementations1 Jul 2019 Matias Valdenegro-Toro

Detecting novel objects without class information is not trivial, as it is difficult to generalize from a small training set.

object-detection Object Detection +1

Deep Neural Networks for Marine Debris Detection in Sonar Images

no code implementations13 May 2019 Matias Valdenegro-Toro

Proper waste disposal and recycling is a must in any sustainable community, and in many coastal areas there is significant water pollution in the form of floating or submerged garbage.

Image Classification

Implementing Noise with Hash functions for Graphics Processing Units

1 code implementation28 Mar 2019 Matias Valdenegro-Toro, Hector Pincheira

We propose a modification to Perlin noise which use computable hash functions instead of textures as lookup tables.


Modeling and Soft-fault Diagnosis of Underwater Thrusters with Recurrent Neural Networks

no code implementations11 Jul 2018 Samy Nascimento, Matias Valdenegro-Toro

Noncritical soft-faults and model deviations are a challenge for Fault Detection and Diagnosis (FDD) of resident Autonomous Underwater Vehicles (AUVs).

Fault Detection General Classification

Improving Predictive Uncertainty Estimation using Dropout -- Hamiltonian Monte Carlo

no code implementations12 May 2018 Diego Vergara, Sergio Hernández, Matias Valdenegro-Toro, Felipe Jorquera

Estimating predictive uncertainty is crucial for many computer vision tasks, from image classification to autonomous driving systems.

Autonomous Driving Bayesian Inference +2

Image Captioning and Classification of Dangerous Situations

no code implementations7 Nov 2017 Octavio Arriaga, Paul Plöger, Matias Valdenegro-Toro

Current robot platforms are being employed to collaborate with humans in a wide range of domestic and industrial tasks.

Anomaly Detection Classification +2

Real-time Convolutional Neural Networks for Emotion and Gender Classification

18 code implementations20 Oct 2017 Octavio Arriaga, Matias Valdenegro-Toro, Paul Plöger

In this paper we propose an implement a general convolutional neural network (CNN) building framework for designing real-time CNNs.

Emotion Classification Face Detection +3

Objectness Scoring and Detection Proposals in Forward-Looking Sonar Images with Convolutional Neural Networks

no code implementations8 Sep 2017 Matias Valdenegro-Toro

In this work we develop a Convolutional Neural Network that can reliably score objectness of image windows in forward-looking sonar images and by thresholding objectness, we generate detection proposals.

Object object-detection +1

Best Practices in Convolutional Networks for Forward-Looking Sonar Image Recognition

no code implementations8 Sep 2017 Matias Valdenegro-Toro

In this work, we evaluate three common decisions that need to be made by a CNN designer, namely the performance of transfer learning, the effect of object/image size and the relation between training set size.

Transfer Learning

Improving Sonar Image Patch Matching via Deep Learning

no code implementations7 Sep 2017 Matias Valdenegro-Toro

Matching sonar images with high accuracy has been a problem for a long time, as sonar images are inherently hard to model due to reflections, noise and viewpoint dependence.

Binary Classification object-detection +3

Real-time convolutional networks for sonar image classification in low-power embedded systems

no code implementations7 Sep 2017 Matias Valdenegro-Toro

Deep Neural Networks have impressive classification performance, but this comes at the expense of significant computational resources at inference time.

General Classification Image Classification

Cannot find the paper you are looking for? You can Submit a new open access paper.