The Experts below are selected from a list of 6999 Experts worldwide ranked by ideXlab platform
Wojciech Samek - One of the best experts on this subject based on the ideXlab platform.
-
GCPR - Identifying individual facial expressions by deconstructing a neural network
Lecture Notes in Computer Science, 2016Co-Authors: Farhad Arbabzadah, Grégoire Montavon, Klaus-robert Müller, Wojciech SamekAbstract:This paper focuses on the problem of explaining predictions of Psychological Attributes such as attractiveness, happiness, confidence and intelligence from face photographs using deep neural networks. Since Psychological Attribute datasets typically suffer from small sample sizes, we apply transfer learning with two base models to avoid overfitting. These models were trained on an age and gender prediction task, respectively. Using a novel explanation method we extract heatmaps that highlight the parts of the image most responsible for the prediction. We further observe that the explanation method provides important insights into the nature of features of the base model, which allow one to assess the aptitude of the base model for a given transfer learning task. Finally, we observe that the multiclass model is more feature rich than its binary counterpart. The experimental evaluation is performed on the 2222 images from the 10k US faces dataset containing Psychological Attribute labels as well as on a subset of KDEF images.
-
Identifying individual facial expressions by deconstructing a neural network
arXiv: Computer Vision and Pattern Recognition, 2016Co-Authors: Farhad Arbabzadah, Grégoire Montavon, Klaus-robert Müller, Wojciech SamekAbstract:This paper focuses on the problem of explaining predictions of Psychological Attributes such as attractiveness, happiness, confidence and intelligence from face photographs using deep neural networks. Since Psychological Attribute datasets typically suffer from small sample sizes, we apply transfer learning with two base models to avoid overfitting. These models were trained on an age and gender prediction task, respectively. Using a novel explanation method we extract heatmaps that highlight the parts of the image most responsible for the prediction. We further observe that the explanation method provides important insights into the nature of features of the base model, which allow one to assess the aptitude of the base model for a given transfer learning task. Finally, we observe that the multiclass model is more feature rich than its binary counterpart. The experimental evaluation is performed on the 2222 images from the 10k US faces dataset containing Psychological Attribute labels as well as on a subset of KDEF images.
Farhad Arbabzadah - One of the best experts on this subject based on the ideXlab platform.
-
GCPR - Identifying individual facial expressions by deconstructing a neural network
Lecture Notes in Computer Science, 2016Co-Authors: Farhad Arbabzadah, Grégoire Montavon, Klaus-robert Müller, Wojciech SamekAbstract:This paper focuses on the problem of explaining predictions of Psychological Attributes such as attractiveness, happiness, confidence and intelligence from face photographs using deep neural networks. Since Psychological Attribute datasets typically suffer from small sample sizes, we apply transfer learning with two base models to avoid overfitting. These models were trained on an age and gender prediction task, respectively. Using a novel explanation method we extract heatmaps that highlight the parts of the image most responsible for the prediction. We further observe that the explanation method provides important insights into the nature of features of the base model, which allow one to assess the aptitude of the base model for a given transfer learning task. Finally, we observe that the multiclass model is more feature rich than its binary counterpart. The experimental evaluation is performed on the 2222 images from the 10k US faces dataset containing Psychological Attribute labels as well as on a subset of KDEF images.
-
Identifying individual facial expressions by deconstructing a neural network
arXiv: Computer Vision and Pattern Recognition, 2016Co-Authors: Farhad Arbabzadah, Grégoire Montavon, Klaus-robert Müller, Wojciech SamekAbstract:This paper focuses on the problem of explaining predictions of Psychological Attributes such as attractiveness, happiness, confidence and intelligence from face photographs using deep neural networks. Since Psychological Attribute datasets typically suffer from small sample sizes, we apply transfer learning with two base models to avoid overfitting. These models were trained on an age and gender prediction task, respectively. Using a novel explanation method we extract heatmaps that highlight the parts of the image most responsible for the prediction. We further observe that the explanation method provides important insights into the nature of features of the base model, which allow one to assess the aptitude of the base model for a given transfer learning task. Finally, we observe that the multiclass model is more feature rich than its binary counterpart. The experimental evaluation is performed on the 2222 images from the 10k US faces dataset containing Psychological Attribute labels as well as on a subset of KDEF images.
Grégoire Montavon - One of the best experts on this subject based on the ideXlab platform.
-
GCPR - Identifying individual facial expressions by deconstructing a neural network
Lecture Notes in Computer Science, 2016Co-Authors: Farhad Arbabzadah, Grégoire Montavon, Klaus-robert Müller, Wojciech SamekAbstract:This paper focuses on the problem of explaining predictions of Psychological Attributes such as attractiveness, happiness, confidence and intelligence from face photographs using deep neural networks. Since Psychological Attribute datasets typically suffer from small sample sizes, we apply transfer learning with two base models to avoid overfitting. These models were trained on an age and gender prediction task, respectively. Using a novel explanation method we extract heatmaps that highlight the parts of the image most responsible for the prediction. We further observe that the explanation method provides important insights into the nature of features of the base model, which allow one to assess the aptitude of the base model for a given transfer learning task. Finally, we observe that the multiclass model is more feature rich than its binary counterpart. The experimental evaluation is performed on the 2222 images from the 10k US faces dataset containing Psychological Attribute labels as well as on a subset of KDEF images.
-
Identifying individual facial expressions by deconstructing a neural network
arXiv: Computer Vision and Pattern Recognition, 2016Co-Authors: Farhad Arbabzadah, Grégoire Montavon, Klaus-robert Müller, Wojciech SamekAbstract:This paper focuses on the problem of explaining predictions of Psychological Attributes such as attractiveness, happiness, confidence and intelligence from face photographs using deep neural networks. Since Psychological Attribute datasets typically suffer from small sample sizes, we apply transfer learning with two base models to avoid overfitting. These models were trained on an age and gender prediction task, respectively. Using a novel explanation method we extract heatmaps that highlight the parts of the image most responsible for the prediction. We further observe that the explanation method provides important insights into the nature of features of the base model, which allow one to assess the aptitude of the base model for a given transfer learning task. Finally, we observe that the multiclass model is more feature rich than its binary counterpart. The experimental evaluation is performed on the 2222 images from the 10k US faces dataset containing Psychological Attribute labels as well as on a subset of KDEF images.
Klaus-robert Müller - One of the best experts on this subject based on the ideXlab platform.
-
GCPR - Identifying individual facial expressions by deconstructing a neural network
Lecture Notes in Computer Science, 2016Co-Authors: Farhad Arbabzadah, Grégoire Montavon, Klaus-robert Müller, Wojciech SamekAbstract:This paper focuses on the problem of explaining predictions of Psychological Attributes such as attractiveness, happiness, confidence and intelligence from face photographs using deep neural networks. Since Psychological Attribute datasets typically suffer from small sample sizes, we apply transfer learning with two base models to avoid overfitting. These models were trained on an age and gender prediction task, respectively. Using a novel explanation method we extract heatmaps that highlight the parts of the image most responsible for the prediction. We further observe that the explanation method provides important insights into the nature of features of the base model, which allow one to assess the aptitude of the base model for a given transfer learning task. Finally, we observe that the multiclass model is more feature rich than its binary counterpart. The experimental evaluation is performed on the 2222 images from the 10k US faces dataset containing Psychological Attribute labels as well as on a subset of KDEF images.
-
Identifying individual facial expressions by deconstructing a neural network
arXiv: Computer Vision and Pattern Recognition, 2016Co-Authors: Farhad Arbabzadah, Grégoire Montavon, Klaus-robert Müller, Wojciech SamekAbstract:This paper focuses on the problem of explaining predictions of Psychological Attributes such as attractiveness, happiness, confidence and intelligence from face photographs using deep neural networks. Since Psychological Attribute datasets typically suffer from small sample sizes, we apply transfer learning with two base models to avoid overfitting. These models were trained on an age and gender prediction task, respectively. Using a novel explanation method we extract heatmaps that highlight the parts of the image most responsible for the prediction. We further observe that the explanation method provides important insights into the nature of features of the base model, which allow one to assess the aptitude of the base model for a given transfer learning task. Finally, we observe that the multiclass model is more feature rich than its binary counterpart. The experimental evaluation is performed on the 2222 images from the 10k US faces dataset containing Psychological Attribute labels as well as on a subset of KDEF images.
Jean-luc Kop - One of the best experts on this subject based on the ideXlab platform.
-
Modeling Psychological Attributes in Psychology – An Epistemological Discussion: Network Analysis vs. Latent Variables
Frontiers in psychology, 2017Co-Authors: Hervé Guyon, Bruno Falissard, Jean-luc KopAbstract:Network Analysis is considered as a new method that challenges Latent Variable models in inferring Psychological Attributes. With Network Analysis, Psychological Attributes are derived from a complex system of components without the need to call on any latent variables. But the ontological status of Psychological Attributes is not adequately defined with Network Analysis, because a Psychological Attribute is both a complex system and a property emerging from this complex system. The aim of this article is to reappraise the legitimacy of latent variable models by engaging in an ontological and epistemological discussion on Psychological Attributes. Psychological Attributes relate to the mental equilibrium of individuals embedded in their social interactions, as robust attractors within complex dynamic processes with emergent properties, distinct from physical entities located in precise areas of the brain. Latent variables thus possess legitimacy, because the emergent properties can be conceptualized and analyzed on the sole basis of their manifestations, without exploring the upstream complex system. However, in opposition with the usual Latent Variable models, this article is in favor of the integration of a dynamic system of manifestations. Latent Variables models and Network Analysis thus appear as complementary approaches. New approaches combining Latent Network Models and Network Residuals are certainly a promising new way to infer Psychological Attributes, placing Psychological Attributes in an inter-subjective dynamic approach. Pragmatism-realism appears as the epistemological framework required if we are to use latent variables as representations of Psychological Attributes.