The DOI matches its Crossref record; 38 of the 38 references it cites were found in a registry or an index; none of the 13 sources its claims cite is checked for support yet.
Tree-CNN: A hierarchical Deep Convolutional Neural Network for incremental learning
Open accessAbstract
Over the past decade, Deep Convolutional Neural Networks (DCNNs) have shown remarkable performance in most computer vision tasks. These tasks traditionally use a fixed dataset, and the model, once trained, is deployed as is. Adding new information to such a model presents a challenge due to complex training issues, such as "catastrophic forgetting", and sensitivity to hyper-parameter tuning. However, in this modern world, data is constantly evolving, and our deep learning models are required to adapt to these changes. In this paper, we propose an adaptive hierarchical network structure composed of DCNNs that can grow and learn as new data becomes available. The network grows in a tree-like fashion to accommodate new classes of data, while preserving the ability to distinguish the previously trained classes. The network organizes the incrementally available data into feature-driven super-classes and improves upon existing hierarchical CNN models by adding the capability of self-growth. The proposed hierarchical model, when compared against fine-tuning a deep network, achieves significant reduction of training effort, while maintaining competitive accuracy on CIFAR-10 and CIFAR-100.
Overview
AI snapshot
Written from the paper’s full text.
TL;DR
Tree-CNN is a hierarchical DCNN that grows tree-like as new classes arrive, assigning new classes to existing, merged, or new child nodes based on softmax likelihood and retraining only affected nodes. On CIFAR-10 and CIFAR-100 it reduces training effort versus fine-tuning a baseline network while maintaining competitive accuracy.
Key findings
- On CIFAR-10, after adding 4 new classes, Tree-CNN reached 86.24% test accuracy with normalized training effort 0.60, versus baseline B:V at 90.51% accuracy and effort 1.00; Tree-CNN used 40% less effort than B:V and 30% less than B:II, while being 3.76% below B:V.
- On CIFAR-100 over 9 incremental stages, Tree-CNN-5/10/20 achieved final test accuracies of 61.57%, 60.46%, and 59.99% (average 69.85%, 69.53%, 68.49%), compared with B:V at 63.05% final / 72.23% average, iCaRL at 49.11% / 64.10%, and LwF at 25.00% / 44.49%. The paper reports Tree-CNNs yield 10% higher accuracy than iCaRL and over 50% higher than LwF.
- Tree-CNNs required lower training effort than fine-tuning cases B:II–B:V; B:I had the lowest effort but the worst accuracy degradation. Tree-CNN-5 grew to 3.4× the size of Network B, Tree-CNN-10 to 2.2×, and Tree-CNN-20 to 1.8×.
- The initial CIFAR-10 Tree-CNN achieved 98.73% root-node testing accuracy, 86% for the “Animals” branch, 94.43% for “Vehicles,” and 89.10% overall.
Methods
- Tree-CNN is composed of DCNNs in a tree: the root classifies inputs into child nodes, branch nodes perform finer classification, and each leaf node is uniquely associated with one class. Incremental learning samples 10% of new-class training images, computes a softmax likelihood matrix, orders classes, then adds each new class to an existing child, merges child nodes, or adds a new child; affected nodes are retrained with old and new data.
- Experiments used CIFAR-10 (6 initial classes, then 4 new classes) and CIFAR-100 (100 classes, 500 training and 100 testing images per class, randomly divided into 10 groups of 10 and introduced over 9 stages). CIFAR-100 used max depth 2, thresholds 0.1, and maxChildren 5, 10, or 20. Baseline Network B was a VGG-like 11-layer network fine-tuned at five depths (B:I–B:V).
- Metrics were testing accuracy and training effort, defined as the sum over nets of total weights × total training samples. Training used MatConvNet, mini-batch SGD with momentum 0.9, dropout, batch normalization, weight decay 0.001, 300 epochs, and learning rate 0.1 for 200 epochs then reduced by 10 every 50 epochs.
Limitations
- The paper states Tree-CNN continues to grow in size over time, and the implications for memory requirements need investigation; it also notes there is no standardized benchmark protocol for incremental learning, so it used a protocol similar to iCaRL.
- The paper states branch nodes have a higher chance of overfitting as dataset size per node shrinks, so more dropout layers were added; it also notes that if the root node misclassifies an input, the final classification is incorrect regardless of the branch node’s output, and old data must be shown during retraining to avoid catastrophic forgetting.
- Evident from the text: evaluation is limited to CIFAR-10 and CIFAR-100; final CIFAR-100 accuracy for Tree-CNN-5 (61.57%) remains below B:V (63.05%) and below HD-CNN (67.38%) and Hertel et al. (67.68%) as listed in Table 7.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 of 13 cited sources checked · 0 supported · 0 flagged
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“The size of Network B remains relatively unchanged, as only additional output nodes are added, which translates to a small fraction of new weights in the final layer.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“However, it suffers the worst accuracy degradation over the 9 learning stages (Fig. 6b).”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The proposed hierarchical model, when compared against fine-tuning a deep network, achieves significant reduction of training effort, while maintaining competitive accuracy on CIFAR-10 and CIFAR-100.”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“At the end of the nine incremental learning stages, certain similar objects grouped together is shown in Fig. 8 for Tree-CNN-10 .”
Strength:plain· not yet calibratedCausal:correlational· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We observe that as new classes are added, the Tree-CNNs grow in size by adding more branches (Fig. 5a).”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We observe that retraining only the FC layers of baseline network ( B:I ) requires the least training effort, however, it gives us the lowest accuracy of 78 : 37% .”
Tests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The root node achieves a testing accuracy of 98 : 73% , while the branch nodes, 'Animals' and 'Vehicles', achieve 86% and 94 : 43% testing accuracy respectively.”
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The Tree-CNNs exhibit a lower training effort than 4 fine-tuning test cases, B:II - B:V .”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Fig. 6b shows that all the networks suffer from some accuracy degradation with increasing number of classes.”
Strength:plain· not yet calibratedCausal:direct· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“An interesting thing to note was similar looking classes, that were also semantically similar, were grouped under the same branches.”
Strength:plain· not yet calibratedCausal:correlational· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passes
- It may state two separate claims as one. 23
- Probably not a substantial claim. 12
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Tree-CNN-5 almost grows 3 : 4 the size of Network B , while Tree-CNN-10 and Tree-CNN-20 reach 2 : 2 and 1 : 8 the baseline size, respectively.”
Strength:plain· not yet calibratedCausal:correlational· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The root node achieves a testing accuracy of 98 : 73% , while the branch nodes, 'Animals' and 'Vehicles', achieve 86% and 94 : 43% testing accuracy respectively.”
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“While there were some groups that had object sharing semantic similarity as well, there were odd groups as well, such as Node 13 as shown in Fig. 8.”
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The three Tree-CNNs perform almost at par with B:IV , and outperform all other variants of network B .”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The Tree-CNN has the second lowest normalized training effort, 40% less than B:V , and 30% less than B:II .”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“At the same time, Tree-CNN had comparable accuracy to B:II and B:III , while just being less than the ideal case B:V by a margin of 3 : 76% .”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
In the paper
“In recent years Deep Convolutional Neural Networks (DCNNs) have emerged as the leading architecture for large scale image classification [1].”
What the sources say
- Not checked yetcites [1] Aljundi, 2016, Expert gate: Lifelong learning with a network of experts, CoRR
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The initial nodes of the Tree-CNN assign the input into coarse super-classes, and as we approach the leaves of the network, finer classification is done.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“While this example is for a two level Tree-CNN, the algorithm can potentially be extended to deeper Tree-CNN models.”
Strength:hedged· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
None of the sources cited for this claim have been checked yet.
In the paper
“However, in these works, the dataset is fixed from the beginning, and prior knowledge of all the classes and their properties is used to build a hierarchical model.”
What the sources say
- Not checked yetcites [22] Sarwar, S. S., Ankit, A., & Roy, K. (2017). Incremental learning in deep convolutional neural networks using partial network sharing, arXiv preprint arXiv:1712.02719.
- Not checked yetcites [23] Sarwar, 2017, Gabor Filter assisted energy efficient fast learning convolutional neural networks
- Not checked yetcites [24] Shmelkov, 2017, Incremental learning of object detectors without catastrophic forgetting
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The decision on how to grow the tree is semi-supervised: the algorithm itself decides how to grow the tree, given the constraints by the user.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesProbably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Each leaf node is uniquely associated to a class and no two leaf nodes have the same class.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✗ Based: failsProbably not a substantial claim. The paper may give no reason why it holds. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Images belonging to the older dataset are required during retraining, but by localizing the change to a small section of the whole network, our method tries to reduce the training effort and complexity.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes? Based: unclearIt may state two separate claims as one.
In the paper
“It has been observed that initial layers of a CNN learn very generic features [17] that has been exploited for transfer learning [18]' [19].”
What the sources say
- Not checked yetcites [17] Panda, 2017, Semantic driven hierarchical learning for energy-efficient image classification
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesProbably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This shows that only retraining the final linear classifier, i.e. the fully connected layer, is not sufficient. We need to train the feature extractors, i.e. convolutional blocks, as well on the new data.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Thus, it can learn with lower training effort than fine-tuning a deep network, while preserving much of the accuracy.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“To compare against the proposed Tree-CNN , we defined a baseline network (Network B ) with a complexity level similar to two stage Tree-CNN .”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✗ Based: failsProbably not a substantial claim. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Such a method of training and classification can be used to hierarchically classify large datasets.”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesProbably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Once the new classes are allotted locations in the tree, supervised gradient descent based training is performed on the modified/new nodes. This saves us from modifying the whole network, and only affected portions of the network require retraining/fine-tuning.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The final leaf nodes, and the distance between them can also be used as a measure of how similar any two images are.”
Strength:plain· not yet calibratedCausal:correlational· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✗ Based: failsProbably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The network organizes the incrementally available data into feature-driven superclasses and improves upon existing hierarchical CNN models by adding the capability of self-growth.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
In the paper
“In 'iCarl' [16], a ResNet-32 is retrained with new data and only 2000 samples of old data (called 'exemplars') at every stage. It is able to recover a good amount performance, compared to 'LwF' but still falls short of state-of-the-art by 18%.”
What the sources say
- Not checked yetcites [16] Panda, 2017, FALCON: Feature driven selective classification for energy-efficient image recognition, IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems source
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The hierarchical node-based learning model of the Tree-CNN attempts to confine the change in the model to a few nodes only. And, in this way, it limits the computation costs of retraining, while using all the previous data.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
In the paper
“Such hierarchical CNN models have been shown to perform at par or even better than standard DCNNs [22].”
What the sources say
- Not checked yetcites [22] Sarwar, S. S., Ankit, A., & Roy, K. (2017). Incremental learning in deep convolutional neural networks using partial network sharing, arXiv preprint arXiv:1712.02719.
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Every class is finally associated with a unique leaf node, hence leaf nodes do not require a look up table.”
Strength:plain· not yet calibratedCausal:direct· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“Training Effort attempts to capture the number of weight updates that happen per training epoch.”
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✗ Based: failsProbably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
In the paper
“For 'LwF', a ResNet-32 [32] is retrained exclusively on new data at every stage. Hence it suffers the most accuracy degradation.”
What the sources say
- Not checked yetcites [32] Yosinski, 2014, How transferable are features in deep neural networks?
Strength:plain· not yet calibratedCausal:direct· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The network grows in a tree-like fashion to accommodate new classes of data, while preserving the ability to distinguish the previously trained classes.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In our work, Tree-CNN starts out as a single root node and generates new hierarchies to accommodate the new classes.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“At every incremental learning stage, the root node is trained on all the available data as it needs to learn to classify all the objects into the new branches.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
In the paper
“The Tree-CNN generates hierarchical grouping of initially unrelated classes, thereby generating a label relation graph out of these classes[33].”
What the sources say
- Not checked yetcites [33]
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“As batch size and number of training epochs is kept the same, the product of the number of weights and the number of training samples used can provide us with the measure of the computation cost of a learning stage.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✗ Based: failsProbably not a substantial claim. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Inspired from hierarchical classifiers, our proposed model, Tree-CNN is composed of multiple nodes connected in a tree-like manner. Each node (except leaf nodes) has a DCNN which is trained to classify the input to the node into one of it's children.”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
None of the sources cited for this claim have been checked yet.
In the paper
“Common features, that are shared between images, have been used previously to build hierarchical classifiers. These features can be grouped semantically, such as in [20], or be feature-driven, such as 'FALCON' [21].”
What the sources say
- Not checked yetcites [20] Russakovsky, 2015, Imagenet large scale visual recognition challenge, International Journal of Computer Vision source
- Not checked yetcites [21] Rusu, A. A., Rabinowitz, N. C., Desjardins, G., Soyer, H., Kirkpatrick, J., & Kavukcuoglu, K., et al. (2016). Progressive neural networks, arXiv preprint arXiv:1606.04671.
Strength:plain· not yet calibratedCausal:direct· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
The paper relies on this without support.
In the real world, data is gathered incrementally over time and all information is not available at once.
“However, in real world, we hardly have all the information at once, and data is, instead, gathered incrementally over time.”
The authors state this as an assumption.
The starting model is already trained to recognize a certain number of objects.
“We start with the assumption that we have a model that is already trained to recognize a certain number of objects.”
The authors state this as an assumption.
The system has access to all the data that has been introduced so far during fine-tuning.
“It is assumed that the system has access to all the data that has been introduced so far.”
The paper relies on this without support.
There is an absence of a standardized benchmark protocol for incremental learning.
“There is an absence of standardized benchmark protocol for incremental learning, which led us to use a benchmarking protocol similar to one used in iCaRL [16].”
The paper relies on this without support.
The dataset available to the user has unique labels assigned to each of its object classes.
“The dataset available to the user will have unique labels assigned to each of it's object classes.”
The paper relies on this without support.
Adding new image classes to a network should be easier than retraining the whole network again with all the classes.
“The motivation of this work stems from the idea that subsequent addition of new image classes to a network should be easier than retraining the whole network again with all the classes.”
The paper relies on this without support.
Catastrophic forgetting is an inherent problem in deep neural networks.
“This is due to the inherent problem of 'catastrophic forgetting' in deep neural networks.”
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Although this method mitigates catastrophic forgetting to a certain extent, it is intuitively a method that consumes more space and is not efficient in training.”
From An Appraisal of Incremental Learning Methods (2020) · cites this paper as [28]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Tree-CNNs do not outperform every Network B variant
MinorThe claim includes B:V, but Table C.9 reports higher accuracy for B:V than for all three Tree-CNNs at 100 classes.
“The three Tree-CNNs perform almost at par with B:IV , and outperform all other variants of network B .”
“| 100 | 31.73 | 48.68 | 54.16 | 60.48 | 63.05 | 61.57 | 60.46 | 59.99 |”
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474
Very Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148
ImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155
Gradient-based learning applied to document recognition
Lecun, Y., Bottou, L., Bengio, Y., & Haffner, P. (1998) · Proceedings of the IEEE · doi:10.1109/5.726791 · Cited by 58,805
ImageNet Large Scale Visual Recognition Challenge
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., et al. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-015-0816-y · Cited by 39,380
(2014) · Cited by 33,419
Girshick, R. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.169 · Cited by 28,105
Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
Ioffe, S., & Szegedy, C. (2015) · arXiv · doi:10.48550/arxiv.1502.03167 · Cited by 27,818
Learning Multiple Layers of Features from Tiny Images
Krizhevsky, A. (2024) · doi:10.57702/zp44cu3g · Cited by 23,889
How transferable are features in deep neural networks?
Yosinski, J., Clune, J., Bengio, Y., & Lipson, H. (2014) · arXiv · doi:10.48550/arxiv.1411.1792 · Cited by 6,061
Flagged first, then by citations
Cited by
Sorted by citations
Items 101–200 of 259
Modular neural network via exploring category hierarchy
Han, W., Zheng, C., Zhang, R., Guo, J., Yang, Q., & Shao, J. (2021) · Information Sciences · doi:10.1016/j.ins.2021.05.032 · Cited by 10
Yi-ren, C., He, K., Hao, B., Weng, Y., & Chen, Z. (2021) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2021.3134650 · Cited by 9
Forecasting short-term methane based on corrected numerical weather prediction outputs
Zhao, S., Wu, L., Xiang, Y., & Zhang, F. (2024) · Journal of Cleaner Production · doi:10.1016/j.jclepro.2024.142500 · Cited by 9
Incremental Learning for Dermatological Imaging Modality Classification
Morgado, A. C., Andrade, C., Teixeira, L. F., & Vasconcelos, M. J. M. (2021) · Journal of Imaging · doi:10.3390/jimaging7090180 · Cited by 9
“[26], Progressive Neural Networks [27], ExpertGate [28], or CWR [29], where explicit architecture modifications were made; or PackNet [30] and Piggyback [31] that use a special mask obtained by network pruning techniques to protect the most relevant weights.”
MethodHierarchical Deep Learning for Human Activity Recognition Integrating Postural Transitions
Tilley, D. J., & Martínez-Hernández, U. (2024) · IEEE Sensors Journal · doi:10.1109/jsen.2024.3491352 · Cited by 8
“Hierarchical convolutional neural networks (CNNs) have been used for image segmentation [17], and hierarchical long short-term memory (LSTM) networks have been examined by [18] using the HAR dataset by University College of Irvine (UCI) [18].”
Mitigate forgetting in few-shot class-incremental learning using different image views
Mazumder, P., & Singh, P. P. (2023) · Neural Networks · doi:10.1016/j.neunet.2023.06.043 · Cited by 8
Ma, J., Chen, Y., Hao, X., Cui, B., & Yang, J. (2024) · Sustainability · doi:10.3390/su16093699 · Cited by 8
Feature Selection-Based Hierarchical Deep Network for Image Classification
He, G., Ji, J., Zhang, H., Xu, Y., & Jianping Fan, B. (2020) · IEEE Access · doi:10.1109/access.2020.2966651 · Cited by 8
“Among them, the hierarchical deep network [12]–[17] combined by deep convolutional neural network [18] and the tree classifier [16] show excellent performance in terms of large-scale object classification.”
BackgroundDRTS Parsing with Structure-Aware Encoding and Decoding
Fu, Q., Zhang, Y., Liu, J., & Zhang, M. (2020) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2020.acl-main.609 · Cited by 8
Intrusion Detection in IoT Network Using Few-Shot Class Incremental Learning
Hosseini, M., & Shi, W. (2024) · Lecture notes in networks and systems · doi:10.1007/978-3-031-54053-0_41 · Cited by 7
Resource-constrained edge-based deep learning for real-time person-identification using foot-pad
Heo, D. H., Park, S. H., & Kang, S. J. (2024) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2024.109290 · Cited by 10
CF-CNN: Coarse-to-Fine Convolutional Neural Network
Park, J., Kim, H., & Paik, J. (2021) · Applied Sciences · doi:10.3390/app11083722 · Cited by 7
“Although this approach may reduce the computational load, the classification accuracy is not preserved [22,24].”
Leveraging joint incremental learning objective with data ensemble for class incremental learning
Mazumder, P., Karim, M. A., Joshi, I., & Singh, P. P. (2023) · Neural Networks · doi:10.1016/j.neunet.2023.01.017 · Cited by 7
Encoding Hierarchical Information in Neural Networks Helps in Subpopulation Shift
Mukherjee, A., Garg, I., & Roy, K. (2023) · IEEE Transactions on Artificial Intelligence · doi:10.1109/tai.2023.3261861 · Cited by 7
“Hierarchical learning for in nonsupervised approaches: Tree-CNN [54] tackles the incremental learning problem where the model expands as a tree to accommodate new classes.”
BackgroundSelf-citationA Framework for Exploring and Modelling Neural Architecture Search Methods
Radiuk, P., & Hrypynska, N. (2020) · Electronic Archive Khmelnitskiy National University (Khmelnitskiy National University) · Cited by 7
Low-power object counting with hierarchical neural networks
Goel, A., Tung, C., Aghajanzadeh, S., Ghodgaonkar, I., Ghosh, S., Thiruvathukal, G. K., & Lu, Y.-H. (2020) · International Symposium on Low Power Electronics and Design (ISLPED) · doi:10.1145/3370748.3406569 · Cited by 7
“only work for image classification (each image has only one object); they cannot be used for object counting (each image has multiple objects) [19, 20].”
BackgroundOnline learning for DBC segmentation of new IGBT samples based on computed laminography imaging
Li, Y., Luo, M., Fei, X., Liu, S., & Wei, C. (2024) · Discover Applied Sciences · doi:10.1007/s42452-024-05807-3 · Cited by 6
Chithra, S., Arunachalaperumal, C., Rajagopal, R., & Meenalochini, P. (2025) · Electrical Engineering · doi:10.1007/s00202-025-02958-3 · Cited by 6
Low-Power Multi-Camera Object Re-Identification using Hierarchical Neural Networks
Goel, A., Tung, C., Hu, X., Wang, H., Davis, J. C., Thiruvathukal, G. K., & Lu, Y.-H. (2021) · International Symposium on Low Power Electronics and Design (ISLPED) · doi:10.1109/islped52811.2021.9502480 · Cited by 6
“[9] show that hierarchies based on visual similarities need to be retrained whenever a previously unseen object is encountered.”
BackgroundMethodTeng, H. (2021) · Computational Intelligence and Neuroscience · doi:10.1155/2021/2794588 · Cited by 6
“Machine learning relies on the process of computer recognition data, which is limited to a certain extent, and requires manpower and material resources to develop specific evaluation indicators for each mental disease [19, 20].”
BackgroundRetractedDetecting network attacks model based on a convolutional neural network
Ali, T. A. J., & Jawhar, M. (2023) · International Journal of Power Electronics and Drive Systems/International Journal of Electrical and Computer Engineering · doi:10.11591/ijece.v13i3.pp3072-3078 · Cited by 6
Kharrat, A., Drira, F., Lebourgeois, F., & Kerautret, B. (2024) · Neurocomputing · doi:10.1016/j.neucom.2024.128217 · Cited by 7
Xi, J., Ersoy, O. K., Fang, J., Cong, M., Wei, X. S., & Wu, T. (2021) · IEEE Access · doi:10.1109/access.2021.3068880 · Cited by 7
A Speech Quality Classifier based on Tree-CNN Algorithm that Considers Network Degradations
Vieira, S. T., Rosa, R. L., & Rodríguez, D. Z. (2020) · Journal of Communications Software and Systems · doi:10.24138/jcomss.v16i2.1032 · Cited by 6
Poonguzhali, P. K., Geetha, P., & Vidhya, R. (2025) · Wireless Networks · doi:10.1007/s11276-024-03893-0 · Cited by 5
“The THDCNN [45] is used for cluster head selection.”
Yao, X., Wu, Y., Xie, G., Zhao, D., & Xia, D. (2025) · Corrosion Engineering Science and Technology The International Journal of Corrosion Processes and Corrosion Control · doi:10.1177/1478422x251387524 · Cited by 6
“80 Interpretability techniques include generating classi fi cation rules through decision path analysis, such as LIME (local interpretable model-agnostic explanations) for mapping local feature contributions, and Shapley values to quantify feature importance.”
Alkahtani, H. K., Ybytayeva, G., Bekarystankyzy, A., & Qahmash, A. (2026) · Scientific Reports · doi:10.1038/s41598-026-39366-x · Cited by 5
“Roy et al. (2020) explored interpretable AI techniques in UAV swarm control by embedding decision trees into deep reinforcement learning networks 39 .”
Devi, B., & Jose, S. E. (2024) · Environment Development and Sustainability · doi:10.1007/s10668-024-05241-w · Cited by 5
“THDCNN is a neural network model used for predicting important data by learning patterns from training input data (Roy et al., 2020).”
Hierarchical neural network with efficient selection inference
Mi, J., Li, N., Huang, K.-Y., Li, W., & Zhou, L. (2023) · Neural Networks · doi:10.1016/j.neunet.2023.02.015 · Cited by 6
Lei, D., Wang, Q., Zhou, F., Kong, J., & Zhou, Z. (2023) · Surface Review and Letters · doi:10.1142/s0218625x23500361 · Cited by 5
Hierarchical Task-Incremental Learning with Feature-Space Initialization Inspired by Neural Collapse
Zhou, Q., Xiang, X., & Ma, J. (2023) · Neural Processing Letters · doi:10.1007/s11063-023-11352-8 · Cited by 5
“[35] proposes a rule-based incremental Tree-CNN structure, and this incremental pattern is consistent with the realistic knowledge expansion.”
ResultShukla, P. K., Alqahtani, A., Dwivedi, A., Alqahtani, N., Shukla, P. K., Alsulami, A. A., Pamucar, D. S., & Šimić, V. (2023) · Neural Computing and Applications · doi:10.1007/s00521-023-09293-3 · Cited by 5
“The lower layers of a hierarchical DL model typically learn fundamental and low-level features from the raw input data, such as edges, textures, or simple patterns [21].”
Autoencoding tree for city generation and applications
Han, W., Wen, C., Chok, L., Tan, Y. L., Chan, S. L., Zhao, H., & Feng, C. (2024) · ISPRS Journal of Photogrammetry and Remote Sensing · doi:10.1016/j.isprsjprs.2024.01.010 · Cited by 5
Reddy, C. R. S. R., Prasanth, B. V., Ogeti, P. S., & Chandra, B. M. (2024) · Journal of Energy Storage · doi:10.1016/j.est.2024.113402 · Cited by 5
A novel database for plant diseases and pests classification
Wang, Q., He, G., Li, F., & Zhang, H. (2020) · IEEE International Conference on Signal Processing, Communications and Computing (ICSPCC) · doi:10.1109/icspcc50002.2020.9259502 · Cited by 6
“However, if a selection is made between several species with high similarity, it will be beneficial for the neural network to continue to explore the different features between similar classes, which can improve the classification accuracy [16–18].”
BackgroundTowards Sustainable Smart City via Resilient Internet of Things
Chui, K. T., de Pablos, P. O., Shen, C.-W., Lytras, M. D., & Vasant, P. (2022) · Contributions to management science · doi:10.1007/978-3-030-85954-1_8 · Cited by 5
Metric-Guided Prototype Learning.Preprint
Garnot, V. S. F., & Landrieu, L. (2020) · arXiv (Cornell University) · Cited by 5
Elevation Correction of a Large-Scale DEM Using ICESat-2 Laser Altimetry Data
Lian, W., Nie, Z., Guo Zhang, A., Li, K., Cao, X., Yu, A., & Li, X. (2025) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2025.3589578 · Cited by 4
Accurate and Robust Roadside 3-D Object Detection Based on Height-Aware Scene Reconstruction
Chen, Y., & Wang, H. (2024) · IEEE Sensors Journal · doi:10.1109/jsen.2024.3444816 · Cited by 4
“Inspired by octree-based voxelization in Tree-CNN [35], we propose a distance-adaptive voxelization method to address this issue.”
MethodMoustakidis, V., Passalis, N., & Tefas, A. (2024) · Pattern Recognition Letters · doi:10.1016/j.patrec.2024.10.005 · Cited by 4
Medical informatization of management system for admission office under convolutional neural network
Wang, G., Li, D., Ge, J., & Wang, T. (2021) · Expert Systems · doi:10.1111/exsy.12780 · Cited by 4
Proposing a Model for Detecting Intrusion Network Attacks Using Machine Learning Techniques
Ali, T. A. J., & Jawhar, M. (2022) · Mağallaẗ al-tarbiyaẗ wa-al-ʻilm · doi:10.33899/edusj.2022.133867.1240 · Cited by 4
“CNNs were first used in image processing applications as a biologically inspired deep learning model for image classification, face recognition, and pattern recognition [24-25].”
MethodPrakash, T. S., Annalakshmi, M., Patnayakuni, S. P., & Shibu, S. (2023) · Journal of Experimental & Theoretical Artificial Intelligence · doi:10.1080/0952813x.2023.2243277 · Cited by 4
Detection-driven 3D masking for efficient object grasping
Li, L., Cherouat, A., Snoussi, H., Hu, R., & Wang, T. (2023) · The International Journal of Advanced Manufacturing Technology · doi:10.1007/s00170-023-12574-9 · Cited by 4
Shallow Hierarchical CNN-LSTM for Activity Recognition to Integrate Postural Transition States
Tilley, D., & Martínez-Hernández, U. (2023) · Proceedings of IEEE Sensors/Proceedings of IEEE Sensors ... · doi:10.1109/sensors56945.2023.10325044 · Cited by 4
“There are wide and varied applications for hierarchical networks: hierarchical CNNs have been used for image segmentation [8], and hierarchical LSTMs have been examined by L. Wang and R. Liu using the HAR dataset by UCI [9].”
MethodIntrusions Detection and Classification Using Deep Learning Approach
Sawadogo, L. M., Bassolé, D., Koala, G., & Sié, O. (2021) · Lecture notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering · doi:10.1007/978-3-030-90556-9_4 · Cited by 4
Han, J., Zhang, T., Li, Y., & Liu, Z. (2022) · International Journal of Machine Learning and Cybernetics · doi:10.1007/s13042-022-01563-1 · Cited by 4
“In addition, the method of dynamically adding model’s structure is also common [40, 41, 45], as shown in Fig.”
MethodSebastian, R. A., & Sebastian, A. M. (2022) · Multimedia Tools and Applications · doi:10.1007/s11042-022-13781-4 · Cited by 4
“Specifically, hierarchical classification [17, 32, 33], semantic embeddings [1, 5, 16, 18, 24], and hierarchy-based losses [6, 11, 24, 38] are extensively used methods to incorporate semantic information for image classification.”
MethodDetecting and Learning the Unknown in Semantic Segmentation
Chan, R., Uhlemeyer, S., Rottmann, M., & Gottschalk, H. (2022) · Springer eBooks · doi:10.1007/978-3-031-01233-4_10 · Cited by 4
Human-inspired Perspectives: A Survey on AI Long-term Memory
He, Z., Lin, W., Zheng, H., Zhang, F., Jones, M. W., Aitchison, L., Xu, X., Liu, M., Kristensson, P. O., & Shen, J. (2024) · arXiv · doi:10.48550/arxiv.2411.00489 · Cited by 3
“For instance, Roy et al. proposed Tree-CNN for handling the introduction of new classes [114].”
AI without borders: The rise of cross-disciplinary machine learning
Wibawa, A. P., Herdianto, R., Handayani, A. N., Utama, A. B. P., Aripriharta, Dwiyanto, F. A., & Dreżewski, R. (2026) · Telematics and Informatics Reports · doi:10.1016/j.teler.2026.100294 · Cited by 4
Directed Acyclic Graph-based Neural Networks for Tunable Low-Power Computer Vision
Goel, A., Tung, C., Eliopoulos, N., Hu, X., Thiruvathukal, G. K., Davis, J. C., & Lu, Y.-H. (2022) · International Symposium on Low Power Electronics and Design (ISLPED) · doi:10.1145/3531437.3539723 · Cited by 3
“In this work, we present methods to provide varying accuracyefficiency tradeoff configurations of a recent low-power DNN architecture, known as the hierarchical DNN [4, 6].”
MethodComplex Facial Expression Recognition Using Deep Knowledge Distillation of Basic Features
Maiden, A., Nguyen, T. V. A., & Nakisa, B. (2025) · International Conference on Digital Image Computing: Techniques and Applications (DICTA) · doi:10.1109/dicta68720.2025.11302420 · Cited by 3
“Overall Accuracy Tree-CNN [23] (0.5107, 0.6437) Fine-tuning [24] (0.6837, 0.7418) LwF [25] (0.5373, 0.6638) TOPIC [26] (0.5268, 0.7168) Deep SLDA [27] (0.5387, 0.7478) REMIND [28] (0.5398, 0.7463) Lucir-CNN [29] (0.5698, 0.7639) PODNet-CNN [30] (0.5991, 0.8163) Lucir [29] /w AANets [31] (0.6414,…”
InfluentialTaxonomy-guided routing in capsule network for hierarchical image classification
Noor, K. T., Luo, W., Robles‐Kelly, A., Zhang, L. Y., & Bouadjenek, M. R. (2025) · Knowledge-Based Systems · doi:10.1016/j.knosys.2025.114444 · Cited by 3
Energy-efficient and Robust Cumulative Training with Net2Net Transformation
Feng, A., & Panda, P. (2020) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn48605.2020.9207451 · Cited by 3
“Partial network sharing strategy for ‘learning without forgetting’ [7] and tree-based efficient neural architectures [24, 25] have also been proposed.”
BackgroundSelf-citationTu, X., Zhang, S., Lai, Z., Xiao, X., & Zhang, L. (2022) · Mobile Information Systems · doi:10.1155/2022/9956753 · Cited by 3
Convolutional Neural Network-Based ECG-Assisted Diagnosis for Coal Workers
Wang, Y., Chen, Z., Tian, S., Zhou, S., Wang, X., Xue, L., & Wu, J. (2022) · International Journal of Environmental Research and Public Health · doi:10.3390/ijerph20010009 · Cited by 3
“Another scholar used CNN to identify the heartbeat and the input-segmented heartbeat, then used the knowledge of CNN to obtain features, and then used more perceptrons to recognize and analyze the input features, which is an end-to-end learning network structure [17,18].”
MethodL 2,1 -Constrained Deep Incremental NMF Approach for SAR Automatic Target Recognition
Cao, C., Chou, R., Zhang, H., Li, X. W., Luo, T., & Liu, B. (2024) · IEEE Geoscience and Remote Sensing Letters · doi:10.1109/lgrs.2024.3361500 · Cited by 3
“An urgent need exists for the proposal of an incremental learning scheme to improve the drawbacks of DNMF-based methods [16], [17], [18].”
BackgroundSilva, I. J., Vilão, C. O., Costa, A. H. R., & Bianchi, R. A. C. (2021) · Journal of Control Automation and Electrical Systems · doi:10.1007/s40313-021-00719-8 · Cited by 3
“…not applied in the robotic mobile agent domain, there is a growing number of research on the multi-class decomposition approach for Deep Learning: Roy et al. (2020) proposed an adaptive hierarchical network structure composed of deep convolutional networks that grows in a tree-like fashion to…”
BackgroundFrosting Weights for Better Continual Training
Zhu, X., Liu, F., Trajcevski, G., & Wang, D. (2019) · International Conference on Machine Learning and Applications · doi:10.1109/icmla.2019.00094 · Cited by 2
“, incremental class learning [13], [19], [23], [38], [41], where different tasks have different classes.”
WITHDRAWN: Optimizing Energy Management in Smart Grid Systems with GJO-THDCNN Hybrid Approach
Chithra, S., Arunachalaperumal, C., Rajagopal, R., & Meenalochini, P. (2025) · Energy · doi:10.1016/j.energy.2025.135386 · Cited by 3
RetractedWu, Y., Wang, M., Min, F., Wang, Q., Zhang, Z.-H., Zhang, H., & Zhou, X. (2025) · IEEE Transactions on Systems Man and Cybernetics Systems · doi:10.1109/tsmc.2025.3552667 · Cited by 6
“Restrictions apply. experts [11], [48], or generated automatically using top-down or bottom-up approaches [49].”
Incremental learning of convolutional neural networks in bioinformatics
Mousser, W., Ouadfel, S., & Ahmed, A. T. (2022) · Elsevier eBooks · doi:10.1016/b978-0-12-820793-2.00001-x · Cited by 2
Use of Sensors for Fall Prediction in Older Persons
Koszalinski, R. S., Tappen, R. M., Ghoraani, B., Vieira, E. R., Marques, O., & Furht, B. (2023) · CIN Computers Informatics Nursing · doi:10.1097/cin.0000000000001052 · Cited by 2
Research on the Short-Term Prediction of Offshore Wind Power Based on Unit Classification
Zhang, J., Liu, X., & Yan, J. (2024) · Electronics · doi:10.3390/electronics13122293 · Cited by 2
Saro: A Simplified Attention-Based Learning-to-Rank Query Optimizer
Wang, J., Cao, S., Sun, C., Zhang, H., Luo, Z., Zhang, B., & Zhao, H. (2024) · IEEE International Conference on e-Business Engineering (ICEBE) · doi:10.1109/icebe62490.2024.00019 · Cited by 2
“Gradient descent keeps fi lters related to execution time stable, and adjusts unrelated fi lters to more useful values, thereby generating more useful features [17].”
Simiyu, D., Muchiri, H., Vikiru, A., Butime, J., & Muti, Z. (2024) · International Conference on Smart Sensors and Application (ICSSA) · doi:10.1109/icssa62312.2024.10788639 · Cited by 2
“CNNs are effective in learning hierarchical representations from input data [21].”
Enhancing query optimization with hybrid ranking model and tree node dependency
Sun, C., Cao, S., Wang, J., Zheng, L., Xie, J., Ruan, C., Wang, H., & Han, Z. (2025) · Service Oriented Computing and Applications · doi:10.1007/s11761-025-00473-8 · Cited by 2
CaptorX: A Class-Adaptive Convolutional Neural Network Reconfiguration Framework
Qin, Z., Yu, F., Xu, Z., Liu, C., & Chen, X. (2021) · IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems · doi:10.1109/tcad.2021.3061520 · Cited by 2
Research on Chest Disease Recognition Based on Deep Hierarchical Learning Algorithm
Li, L., Long, Y., Huang, B., Chen, Z., Liu, Z., & Yang, Z. (2022) · Journal of Healthcare Engineering · doi:10.1155/2022/6996444 · Cited by 2
RetractedClass-Incremental Learning on Video-Based Action Recognition by Distillation of Various Knowledge
Maraghi, V. O., & Karim Faez, A. (2022) · Computational Intelligence and Neuroscience · doi:10.1155/2022/4879942 · Cited by 2
“An adaptive hierarchical network composed of DCNNs is proposed for incremental learning that grows and learns as new data becomes available [25].”
BackgroundPrototype Representation Expansion in Incremental Learning
Mao, K., Luo, Y., Ren, Y., & Wang, R. (2023) · Neural Processing Letters · doi:10.1007/s11063-023-11317-x · Cited by 2
“In Progressive neural networks (PNN) [30] and Tree-CNN [28], the parameters that have been trained will be fixed and the networks will continue to expand.”
BackgroundLifelong Machine Learning for Regional-Based Image Classification in Open Datasets
Alyami, H., Alharbi, A., & Uddin, M. I. (2020) · Symmetry · doi:10.3390/sym12122094 · Cited by 2
“In this section, we devised a technique based on Tree-CNN [28], where models are able to share information gained from different datasets and hence we are able to reduce the gap of geo-diversity in open datasets.”
MethodGraph Tree Neural NetworksPreprint
Kim, S., Jang, J., Jung, H., & Kim, H. (2021) · arXiv (Cornell University) · Cited by 2
Intrusion Detection in IoT Network using Few-Shot Class Incremental Learning
Hosseini, S. (2023) · doi:10.22215/etd/2023-15668 · Cited by 2
Scene Text Detection and Recognition Based on Iterative Correction
Xiong, L. W., Gui, Z., Ou, Y., & Xu, W. (2022) · doi:10.1145/3556267.3556269 · Cited by 1
Deep Lifelong Learning Optimization Algorithm in Dense Region Fusion
Zhang, L., Ding, F., Xiang, S., Qin, Z., Chang, Z., & Wang, H. (2023) · Applied Sciences · doi:10.3390/app13137549 · Cited by 1
Zitouni, A., Falek, L., Amrouche, A., Dahou, B., & Abbas, M. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-18355-0 · Cited by 1
Seethapathi, M., Robert, J. B., Rajesh, P., & Shajin, F. H. (2024) · International Journal of Pavement Engineering · doi:10.1080/10298436.2024.2393312 · Cited by 1
“By identifying and connecting patterns from the training input data samples, the neural network technique is utilised to forecast the useful data (Roy et al .”
Segmentation-Based CCTA Image Analysis to Support the Diagnosis of Coronary Artery Disease
Settanni, G., Vaira, R., Spedalieri, M., Colatruglio, M., Savino, N., & Galiano, A. M. (2024) · IEEE Canada International Humanitarian Technology Conference · doi:10.1109/ihtc61819.2024.10855144 · Cited by 1
“This architecture has been proposed to address the limitations of the traditional U-Net, especially when it comes to handling high-resolution images [8].”
Prediction of Coatings Hardness and Wear Rate by a Continual Learning Model with Dynamic Structure
Wang, Q., Lei, D., Kong, J., & Zhou, Z. (2025) · Journal of Materials Engineering and Performance · doi:10.1007/s11665-025-11308-0 · Cited by 1
O’MINE: A Novel Collaborative DDoS Detection Mechanism for Programmable Data-Planes
Bardhi, E., Ji, C., Imran, A., Shahbaz, M., Lazzeretti, R., Conti, M., & Kuipers, F. A. (2025) · IEEE European Symposium on Security and Privacy · doi:10.1109/eurosp63326.2025.00049 · Cited by 1
“Ways of mitigating the data drift concept that we might adopt for O’ MINE in the future encompass transfer learning [64] and online retraining on unsupervised new data [65], [66].”
Kundu, S., Roy, N., Talukdar, R., Das, S., Mukhopadhyay, S., Mallik, B. B., Majumdar, S., Kirkley, S. E., & Basu, S. (2025) · International Conference on Information Technology · doi:10.1109/incit66780.2025.11276115 · Cited by 1
“Roy et al. [22] developed a Deep Convolutional Neural Network (DCNN) endowed with a hierarchical topology so it might be adaptable with newly inserted classes of data.”
A Deep Learning Approach To Analyze The Sentiment Of Online Game Users
Wahana, A., Zulfikar, W. B., Wildiansyah, W. N., Atmadja, A. R., Ramdania, D. R., & Subaeki, B. (2021) · International Conference on Wireless and Telematics (ICWT) · doi:10.1109/icwt52862.2021.9678416 · Cited by 1
An Effective Ensemble Model Related to Incremental Learning in Neural Machine Translation
Shi, P. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-30105-6_10 · Cited by 1
Pretrained back propagation based adaptive resonance theory network for adaptive learning
Zhang, C., Jiang, C., & Xu, Q. (2023) · Journal of Algorithms & Computational Technology · doi:10.1177/17483026231205009 · Cited by 3
“33 From the perspective of network structure, the former uses structure search to adapt to new tasks quickly, while the latter uses CNN for effective feature extraction and combines with a decision tree network for adaptive growth.”
MethodMulti-channel Episodic Memory Building using Recurrent Kernel Machine
Naseer, S. A., Dawood, F., & Zubair, M. (2023) · International Multi-Topic Conference · doi:10.1109/inmic60434.2023.10466071 · Cited by 1
“In [25] this research, they proposed CNNs tree like structure, that grows hierarchically, and each node presented classes in tree.”
MethodExperimental study of rehearsal-based incremental classification of document streams
Malik, U., Visani, M., Sidère, N., Coustaty, M., & Joseph, A. (2024) · International Journal on Document Analysis and Recognition (IJDAR) · doi:10.1007/s10032-024-00467-w · Cited by 1
“Roy et al. [26] created a hierarchical framework with a tree structure.”
Hierarchical multi-granularity classification based on bidirectional knowledge transfer
Jiang, J., Yang, J., Zhang, W., & Zhang, H. (2024) · Multimedia Systems · doi:10.1007/s00530-024-01412-x · Cited by 1
“Currently, most existing studies [11–13] use hierarchical trees for hierarchical multi-granularity classification.”
Ramasamy, K., Manoharan, M., Narayanasamy, P., & Williams, R. B. (2024) · Environment Development and Sustainability · doi:10.1007/s10668-024-05507-3 · Cited by 1
“The THDCNN model uses learning and correlation patterns from training data to forecast valuable data, such as liquid level prediction (Roy et al., 2020).”
ILKD: An Incremental Learning and Knowledge Distillation Framework for Coronavirus Risk Prediction
Shen, J., Shen, A., Huang, H., Hou, Y., & Kou, Z. (2024) · Journal of Physics Conference Series · doi:10.1088/1742-6596/2868/1/012025 · Cited by 1
Influencer Ranking Framework Using TH-DCNN for influence maximization
Shelke, V., & Jadhav, A. (2025) · Procedia Computer Science · doi:10.1016/j.procs.2025.01.018 · Cited by 1
Dual prototypes for adaptive pre-trained model in class-incremental learning
Xu, Z., Yang, S., Xu, B., Shen, F., & Zhao, J. (2025) · Neural Networks · doi:10.1016/j.neunet.2025.108389 · Cited by 1
He, G., Zhang, Q., Zhang, H., Xu, Y., & Jianping Fan, B. (2020) · Multimedia Tools and Applications · doi:10.1007/s11042-020-09090-3 · Cited by 1
“Adding these relationships to the model can effectively improve image classification; The literature [10, 15, 19, 26, 29, 34, 40, 55] constructs a two-level taxonomy, with is a rough way to organize a large number of categories to form a concept ontology; Zhang Xiaofan et al. first paid attention…”
MethodDeep Neural Networks: Incremental Learning
Murthy, G. R., Bairaju, M. L., Prasanna, G. J., Vankam, V. S., & Jagannadan, M. (2020) · Advances in intelligent systems and computing · doi:10.1007/978-3-030-55180-3_46 · Cited by 1
Quality Control on Manufacturing Computer Keyboards Using Multilevel Deep Neural Networks
Huang, X., & Jing Ren, A. (2020) · IEEE International Conference on Control Science and Systems Engineering (ICCSSE) · doi:10.1109/iccsse50399.2020.9171988 · Cited by 1
“In [15], a hierarchical tree structure of Convolutional Neural Network (CNN) is designed for incremental learning.”
BackgroundWang, K., Deng, J., Xu, L., Tang, C., Pei, Z., & Wang, H. (2020) · Chinese Control Conference (CCC) · doi:10.23919/ccc50068.2020.9189086 · Cited by 1
“CNN [7] [2] [8] which is the high-performance algorithms has achieved a mentionable result of a 0.4 % error rate on the MNIST handwritten character recognition dataset, using a fairly simple CNN and plus elastic training image deformations to increase the training data size.”
MethodInfluentialLiu, X., Cheng, J. T., & Zhang, Q. (2021) · IEEE Access · doi:10.1109/access.2021.3062231 · Cited by 1
“…information, we use four branch streams, pertaining to the global semantics of the sentence sequences based on the Tree LSTM [33] and Tree CNN [34] (denoted by f sc ( R ) in FIGURE 2), local semantics of the sentence sequences based on the word embedding MLP and Tree CNN (denoted by f wc ( R…”
MethodMirrashid, A., & Shirazi, A. A. B. (2021) · Computational Intelligence · doi:10.1111/coin.12472 · Cited by 1
Anytime Inference with Distilled Hierarchical Neural Ensembles
Ruiz, A., & Verbeek, J. J. (2021) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v35i11.17140 · Cited by 1
“This is a significant performance boost given the difficulty and large-scale nature of ImageNet.”
MethodInfluential
Metadata
- Authors
- Roy, D., Panda, P., & Roy, K.
- Affiliations
- Purdue University West Lafayette (US)
- Corresponding author
- Roy, Deboleena
- Venue
- Neural Networks
- Publisher
- Elsevier BV
- Published
- 2020-01
- Volume
- 121
- Pages
- 148-160
- ISSN
- 0893-6080 (ISSN-L)
- OpenAlex
- W2804177786
- Semantic Scholar
- e811c76173d4b0e478d3ab5965358a6377dd48ce
- Cited by
- 257
- References
- 37
- Citations by year
- 2026: 12 · 2025: 29 · 2024: 39 · 2023: 44 · 2022: 41 · 2021: 53 · 2020: 30
- FWCI
- 15.11
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Licence
- Locations
- Neural Networks (acceptedVersion)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- Semiconductor Research Corporation
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Data Stream Mining Techniques
- Keywords
- incremental learning, catastrophic forgetting, hierarchical convolutional neural network, class-incremental learning, adaptive network structure
- MeSH terms
- Deep Learning, Animals, Humans, Pattern Recognition, Automated, Photic Stimulation, Neural Networks, Computer
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-08, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-08 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
