Regular Polytope Networks

Pernici, Federico; Bruni, Matteo; Baecchi, Claudio; Bimbo, Alberto Del

doi:10.1109/tnnls.2021.3056762

Cited by 11 publications

(4 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the continual re-identification problem, [63] shows that performance does not decrease during learning of multiple datasets. This evidence opens up also to recent methodologies for learning the features so-called compatible [67], based on fixed classifiers [58][59][60], in which the re-indexing of the gallery is no longer necessary when updating the feature representation.…”

Section: Fine Grained Visual Categorizationmentioning

confidence: 88%

Fine-Grained Adversarial Semi-Supervised Learning

Mugnai

Pernici

Turchini

et al. 2022

ACM Trans. Multimedia Comput. Commun. Appl.

Self Cite

View full text Add to dashboard Cite

In this article, we exploit Semi-Supervised Learning ( SSL ) to increase the amount of training data to improve the performance of Fine-Grained Visual Categorization ( FGVC ). This problem has not been investigated in the past in spite of prohibitive annotation costs that FGVC requires. Our approach leverages unlabeled data with an adversarial optimization strategy in which the internal features representation is obtained with a second-order pooling model. This combination allows one to back-propagate the information of the parts, represented by second-order pooling, onto unlabeled data in an adversarial training setting. We demonstrate the effectiveness of the combined use by conducting experiments on six state-of-the-art fine-grained datasets, which include Aircrafts, Stanford Cars, CUB-200-2011, Oxford Flowers, Stanford Dogs, and the recent Semi-Supervised iNaturalist-Aves. Experimental results clearly show that our proposed method has better performance than the only previous approach that examined this problem; it also obtained higher classification accuracy with respect to the supervised learning methods with which we compared.

show abstract

Section: Fine Grained Visual Categorizationmentioning

confidence: 88%

Fine-Grained Adversarial Semi-Supervised Learning

Mugnai

Pernici

Turchini

et al. 2022

ACM Trans. Multimedia Comput. Commun. Appl.

Self Cite

View full text Add to dashboard Cite

show abstract

“…For the continual re-identification problem, [63] shows that performance does not decrease during learning from multiple datasets. This evidence opens up also to recent methodologies for learning the features so-called compatible [67], based on the concept of fixed classifiers [58][59][60], in which the re-indexing of the gallery is no longer necessary when upgrading the feature representation.…”

Section: Fine Grained Visual Categorizationmentioning

confidence: 91%

Fine-Grained Adversarial Semi-supervised Learning

Mugnai¹,

Pernici²,

Turchini³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

In this paper we exploit Semi-Supervised Learning (SSL) to increase the amount of training data to improve the performance of Fine-Grained Visual Categorization (FGVC). This problem has not been investigated in the past in spite of prohibitive annotation costs that FGVC requires. Our approach leverages unlabeled data with an adversarial optimization strategy in which the internal features representation is obtained with a second-order pooling model. This combination allows to back-propagate the information of the parts, represented by second-order pooling, onto unlabeled data in an adversarial training setting. We demonstrate the effectiveness of the combined use by conducting experiments on six state-of-the-art fine-grained datasets, which include Aircrafts, Stanford Cars, CUB-200-2011, Oxford Flowers, Stanford Dogs, and the recent Semi-Supervised iNaturalist-Aves. Experimental results clearly show that our proposed method has better performance than the only previous approach that examined this problem; it also obtained higher classification accuracy with respect to the supervised learning methods with which we compared.

show abstract

“…Recently, it has been argued that the weights of the output layer can be fixed as non-trainable with no loss in accuracy [31], [32]. The main intuition is that during the training, both input features and weight vectors align simultaneously but fixing output layer weights still is able learn by adapting to input features.…”

Section: Neural Network Architecturementioning

confidence: 99%

“…Therefore, we fix the weight matrix 𝑊 𝐿 of the output layer to be an the Hadamard approximation of an orthonormal projection matrix 𝑃 ∈ 𝐑 𝑛 𝐿−1 ×𝐾 where 𝑛 𝐿−1 is the size of the last hidden layer. The output 𝑦 of the network is given by [32]:…”

Section: Neural Network Architecturementioning

confidence: 99%

Sparse Deep Neural Network for Encoding and Decoding the Structural Connectome

Singh,

Gupta,

Rajapakse

2024

IEEE J. Transl. Eng. Health Med.

View full text Add to dashboard Cite

Brain state classification by applying deep learning techniques on neuroimaging data has become a recent topic of research. However, unlike domains where the data is low dimensional or there are large number of available training samples, neuroimaging data is high dimensional and has few training samples. To tackle these issues, we present a sparse feedforward deep neural architecture for encoding and decoding the structural connectome of the human brain. We use a sparsely connected element-wise multiplication as the first hidden layer and a fixed transform layer as the output layer. The number of trainable parameters and the training time is significantly reduced compared to feedforward networks. We demonstrate superior performance of this architecture in encoding the structural connectome implicated in Alzheimer’s disease (AD) and Parkinson’s disease (PD) from DTI brain scans. For decoding, we propose recursive feature elimination (RFE) algorithm based on DeepLIFT, layer-wise relevance propagation (LRP), and Integrated Gradients (IG) algorithms to remove irrelevant features and thereby identify key biomarkers associated with AD and PD. We show that the proposed architecture reduces 45.1% and 47.1% of the trainable parameters compared to a feedforward DNN with an increase in accuracy by 2.6 % and 3.1% for cognitively normal (CN) vs AD and CN vs PD classification, respectively. We also show that the proposed RFE method leads to a further increase in accuracy by 2.1% and 4% for CN vs AD and CN vs PD classification, while removing approximately 90% to 95% irrelevant features. Furthermore, we argue that the biomarkers (i.e., key brain regions and connections) identified are consistent with previous literature. We show that relevancy score-based methods can yield high discriminative power and are suitable for brain decoding. We also show that the proposed approach led to a reduction in the number of trainable network parameters, an increase in classification accuracy, and a detection of brain connections and regions that were consistent with earlier studies.

show abstract

Regular Polytope Networks

Cited by 11 publications

References 49 publications

Fine-Grained Adversarial Semi-Supervised Learning

Fine-Grained Adversarial Semi-Supervised Learning

Fine-Grained Adversarial Semi-supervised Learning

Sparse Deep Neural Network for Encoding and Decoding the Structural Connectome

Contact Info

Product

Resources

About