Temporal Neural Networks (TNNs) are spiking neural networks that use time as a resource to represent and process information, similar to the mammalian neocortex. In contrast to compute-intensive deep neural networks that employ separate training and inference phases, TNNs are capable of extremely efficient online incremental/continual learning and are excellent candidates for building edge-native sensory processing units. This work proposes a microarchitecture framework for implementing TNNs using standard CMOS. Gate-level implementations of three key building blocks are presented: 1) multi-synapse neurons, 2) multi-neuron columns, and 3) unsupervised and supervised online learning algorithms based on Spike Timing Dependent Plasticity (STDP). The proposed microarchitecture is embodied in a set of characteristic scaling equations for assessing the gate count, area, delay and power for any TNN design. Post-synthesis results (in 45nm CMOS) for the proposed designs are presented, and their online incremental learning capability is demonstrated.
The existing low-memory BLS implementation proposed recently avoids the need for storing and inverting large matrices, to achieve efficient usage of memories. However, the existing low-memory BLS implementation sacrifices the testing accuracy as a price for efficient usage of memories, since it can no longer obtain the generalized inverse or ridge solution for the output weights during incremental learning, and it cannot work under the very small ridge parameter that is utilized in the original BLS. Accordingly, it is required to develop the low-memory BLS implementations, which can work under very small ridge parameters and compute the generalized inverse or ridge solution for the output weights in the process of incremental learning. In this paper, firstly we propose the low-memory implementations for the recently proposed recursive and square-root BLS algorithms on added inputs and the recently proposed squareroot BLS algorithm on added nodes, by simply processing a batch of inputs or nodes in each recursion. Since the recursive BLS implementation includes the recursive updates of the inverse matrix that may introduce numerical instabilities after a large number of iterations, and needs the extra computational load to decompose the inverse matrix into the Cholesky factor when cooperating with the proposed low-memory implementation of the square-root BLS algorithm on added nodes, we only improve the low-memory implementations of the square-root BLS algorithms on added inputs and nodes, to propose the full lowmemory implementation of the square-root BLS algorithm. All the proposed low-memory BLS implementations compute the ridge solution for the output weights in the process of incremental learning, and most of them can work under very small ridge parameters.
Deep Learning (DL) has been utilized pervasively in the Internet of Things (IoT). One typical application of DL in IoT is device identification from wireless signals, namely Non-cryptographic Device Identification (NDI). However, learning components in NDI systems have to evolve to adapt to operational variations, such a paradigm is termed as Incremental Learning (IL). Various IL algorithms have been proposed and many of them require dedicated space to store the increasing amount of historical data, and therefore, they are not suitable for IoT or mobile applications. However, conventional IL schemes can not provide satisfying performance when historical data are not available. In this paper, we address the IL problem in NDI from a new perspective, firstly, we provide a new metric to measure the degree of topological maturity of DNN models from the degree of conflict of class-specific fingerprints. We discover that an important cause for performance degradation in IL enabled NDI is owing to the conflict of devices' fingerprints. Second, we also show that the conventional IL schemes can lead to low topological maturity of DNN models in NDI systems. Thirdly, we propose a new Channel Separation Enabled Incremental Learning (CSIL) scheme without using historical data, in which our strategy can automatically separate devices' fingerprints in different learning stages and avoid potential conflict. Finally, We evaluated the effectiveness of the proposed framework using real data from ADS-B (Automatic Dependent Surveillance-Broadcast), an application of IoT in aviation. The proposed framework has the potential to be applied to accurate identification of IoT devices in a variety of IoT applications and services. Data and code available at IEEE Dataport (DOI: 10.21227/1bxc-ke87) and \url{https://github.com/pcwhy/CSIL}}
Predefined evenly-distributed class centroids (PEDCC) can be widely used in models and algorithms of pattern classification, such as CNN classifiers, classification autoencoders, clustering, and semi-supervised learning, etc. Its basic idea is to predefine the class centers, which are evenly-distributed on the unit hypersphere in feature space, to maximize the inter-class distance. The previous method of generating PEDCC uses an iterative algorithm based on a charge model, that is, the initial values of various centers (charge positions) are randomly set from the normal distribution, and the charge positions are updated iteratively with the help of the repulsive force between charges of the same polarity. The class centers generated by the algorithm will produce some errors with the theoretically evenly-distributed points, and the generation time will be longer. This paper takes advantage of regular polyhedron in high-dimensional space and the evenly distribution of points on the n dimensional hypersphere to generate PEDCC mathematically. Then, we discussed the basic and extensive characteristics of the frames formed by PEDCC. Finally, experiments show that new algorithm is not only faster than the iterative method, but also more accurate in position. The mathematical analysis and experimental results of this paper can provide a theoretical tool for using PEDCC to solve the key problems in the field of pattern recognition, such as interpretable supervised/unsupervised learning, incremental learning, uncertainty analysis and so on.
In the last years, deep learning has dramatically improved the performances in a variety of medical image analysis applications. Among different types of deep learning models, convolutional neural networks have been among the most successful and they have been used in many applications in medical imaging. Training deep convolutional neural networks often requires large amounts of image data to generalize well to new unseen images. It is often time-consuming and expensive to collect large amounts of data in the medical image domain due to expensive imaging systems, and the need for experts to manually make ground truth annotations. A potential problem arises if new structures are added when a decision support system is already deployed and in use. Since the field of radiation therapy is constantly developing, the new structures would also have to be covered by the decision support system. In the present work, we propose a novel loss function, that adapts to the available data in order to utilize all available data, even when some have missing annotations. We demonstrate that the proposed loss function also works well in an incremental learning setting, where it can automatically incorporate new structures as they appear. Experiments on a large in-house data set show that the proposed method performs on par with baseline models, while greatly reducing the training time.
Incremental learning aims to enable machine learning models to continuously acquire new knowledge given new classes, while maintaining the knowledge already learned for old classes. Saving a subset of training samples of previously seen classes in the memory and replaying them during new training phases is proven to be an efficient and effective way to fulfil this aim. It is evident that the larger number of exemplars the model inherits the better performance it can achieve. However, finding a trade-off between the model performance and the number of samples to save for each class is still an open problem for replay-based incremental learning and is increasingly desirable for real-life applications. In this paper, we approach this open problem by tapping into a two-step compression approach. The first step is a lossy compression, we propose to encode input images and save their discrete latent representations in the form of codes that are learned using a hierarchical Vector Quantised Variational Autoencoder (VQ-VAE). In the second step, we further compress codes losslessly by learning a hierarchical latent variable model with bits-back asymmetric numeral systems (BB-ANS). To compensate for the information lost in the first step compression, we introduce an Information Back (IB) mechanism that utilizes real exemplars for a contrastive learning loss to regularize the training of a classifier. By maintaining all seen exemplars' representations in the format of `codes', Discrete Representation Replay (DRR) outperforms the state-of-art method on CIFAR-100 by a margin of 4% accuracy with a much less memory cost required for saving samples. Incorporated with IB and saving a small set of old raw exemplars as well, the accuracy of DRR can be further improved by 2% accuracy.
Incrementally training deep neural networks to recognize new classes is a challenging problem. Most existing class-incremental learning methods store data or use generative replay, both of which have drawbacks, while 'rehearsal-free' alternatives such as parameter regularization or bias-correction methods do not consistently achieve high performance. Here, we put forward a new strategy for class-incremental learning: generative classification. Rather than directly learning the conditional distribution p(y|x), our proposal is to learn the joint distribution p(x,y), factorized as p(x|y)p(y), and to perform classification using Bayes' rule. As a proof-of-principle, here we implement this strategy by training a variational autoencoder for each class to be learned and by using importance sampling to estimate the likelihoods p(x|y). This simple approach performs very well on a diverse set of continual learning benchmarks, outperforming generative replay and other existing baselines that do not store data.
We present an approach for continual learning (CL) that is based on fully probabilistic (or generative) models of machine learning. In contrast to, e.g., GANs that are "generative" in the sense that they can generate samples, fully probabilistic models aim at modeling the data distribution directly. Consequently, they provide functionalities that are highly relevant for continual learning, such as density estimation (outlier detection) and sample generation. As a concrete realization of generative continual learning, we propose Gaussian Mixture Replay (GMR). GMR is a pseudo-rehearsal approach using a Gaussian Mixture Model (GMM) instance for both generator and classifier functionalities. Relying on the MNIST, FashionMNIST and Devanagari benchmarks, we first demonstrate unsupervised task boundary detection by GMM density estimation, which we also use to reject untypical generated samples. In addition, we show that GMR is capable of class-conditional sampling in the way of a cGAN. Lastly, we verify that GMR, despite its simple structure, achieves state-of-the-art performance on common class-incremental learning problems at very competitive time and memory complexity.
Neural networks notoriously suffer from the problem of catastrophic forgetting, the phenomenon of forgetting the past knowledge when acquiring new knowledge. Overcoming catastrophic forgetting is of significant importance to emulate the process of "incremental learning", where the model is capable of learning from sequential experience in an efficient and robust way. State-of-the-art techniques for incremental learning make use of knowledge distillation towards preventing catastrophic forgetting. Therein, one updates the network while ensuring that the network's responses to previously seen concepts remain stable throughout updates. This in practice is done by minimizing the dissimilarity between current and previous responses of the network one way or another. Our work contributes a novel method to the arsenal of distillation techniques. In contrast to the previous state of the art, we propose to firstly construct low-dimensional manifolds for previous and current responses and minimize the dissimilarity between the responses along the geodesic connecting the manifolds. This induces a more formidable knowledge distillation with smooth properties which preserves the past knowledge more efficiently as observed by our comprehensive empirical study.
Continual learning aims to learn new tasks incrementally using less computation and memory resources instead of retraining the model from scratch whenever new task arrives. However, existing approaches are designed in supervised fashion assuming all data from new tasks have been manually annotated, which are not practical for many real-life applications. In this work, we propose to use pseudo label instead of the ground truth to make continual learning feasible in unsupervised mode. The pseudo labels of new data are obtained by applying global clustering algorithm and we propose to use the model updated from last incremental step as the feature extractor. Due to the scarcity of existing work, we introduce a new benchmark experimental protocol for unsupervised continual learning of image classification task under class-incremental setting where no class label is provided for each incremental learning step. Our method is evaluated on the CIFAR-100 and ImageNet (ILSVRC) datasets by incorporating the pseudo label with various existing supervised approaches and show promising results in unsupervised scenario.
In this paper we analyze the classification performance of neural network structures without parametric inference. Making use of neural architecture search, we empirically demonstrate that it is possible to find random weight architectures, a deep prior, that enables a linear classification to perform on par with fully trained deep counterparts. Through ablation experiments, we exclude the possibility of winning a weight initialization lottery and confirm that suitable deep priors do not require additional inference. In an extension to continual learning, we investigate the possibility of catastrophic interference free incremental learning. Under the assumption of classes originating from the same data distribution, a deep prior found on only a subset of classes is shown to allow discrimination of further classes through training of a simple linear classifier.
Recent advances in unsupervised domain adaptation have significantly improved the recognition accuracy of CNNs by alleviating the domain shift between (labeled) source and (unlabeled) target data distributions. While the problem of single-target domain adaptation (STDA) for object detection has recently received much attention, multi-target domain adaptation (MTDA) remains largely unexplored, despite its practical relevance in several real-world applications, such as multi-camera video surveillance. Compared to the STDA problem that may involve large domain shifts between complex source and target distributions, MTDA faces additional challenges, most notably the computational requirements and catastrophic forgetting of previously-learned targets, which can depend on the order of target adaptations. STDA for detection can be applied to MTDA by adapting one model per target, or one common model with a mixture of data from target domains. However, these approaches are either costly or inaccurate. The only state-of-art MTDA method specialized for detection learns targets incrementally, one target at a time, and mitigates the loss of knowledge by using a duplicated detection model for knowledge distillation, which is computationally expensive and does not scale well to many domains. In this paper, we introduce an efficient approach for incremental learning that generalizes well to multiple target domains. Our MTDA approach is more suitable for real-world applications since it allows updating the detection model incrementally, without storing data from previous-learned target domains, nor retraining when a new target domain becomes available. Our proposed method, MTDA-DTM, achieved the highest level of detection accuracy compared against state-of-the-art approaches on several MTDA detection benchmarks and Wildtrack, a benchmark for multi-camera pedestrian detection.
In domains where users tend to develop long-term preferences that do not change too frequently, the stability of recommendations is an important factor of the perceived quality of a recommender system. In such cases, unstable recommendations may lead to poor personalization experience and distrust, driving users away from a recommendation service. We propose an incremental learning scheme that mitigates such problems through the dynamic modeling approach. It incorporates a generalized matrix form of a partial differential equation integrator that yields a dynamic low-rank approximation of time-dependent matrices representing user preferences. The scheme allows extending the famous PureSVD approach to time-aware settings and significantly improves its stability without sacrificing the accuracy in standard top-$n$ recommendations tasks.
While many works on Continual Learning have shown promising results for mitigating catastrophic forgetting, they have relied on supervised training. To successfully learn in a label-agnostic incremental setting, a model must distinguish between learned and novel classes to properly include samples for training. We introduce a novelty detection method that leverages network confusion caused by training incoming data as a new class. We found that incorporating a class-imbalance during this detection method substantially enhances performance. The effectiveness of our approach is demonstrated across a set of image classification benchmarks: MNIST, SVHN, CIFAR-10, CIFAR-100, and CRIB.
Few-shot class-incremental learning (FSCIL) aims to design machine learning algorithms that can continually learn new concepts from a few data points, without forgetting knowledge of old classes. The difficulty lies in that limited data from new classes not only lead to significant overfitting issues but also exacerbate the notorious catastrophic forgetting problems. Moreover, as training data come in sequence in FSCIL, the learned classifier can only provide discriminative information in individual sessions, while FSCIL requires all classes to be involved for evaluation. In this paper, we address the FSCIL problem from two aspects. First, we adopt a simple but effective decoupled learning strategy of representations and classifiers that only the classifiers are updated in each incremental session, which avoids knowledge forgetting in the representations. By doing so, we demonstrate that a pre-trained backbone plus a non-parametric class mean classifier can beat state-of-the-art methods. Second, to make the classifiers learned on individual sessions applicable to all classes, we propose a Continually Evolved Classifier (CEC) that employs a graph model to propagate context information between classifiers for adaptation. To enable the learning of CEC, we design a pseudo incremental learning paradigm that episodically constructs a pseudo incremental learning task to optimize the graph parameters by sampling data from the base dataset. Experiments on three popular benchmark datasets, including CIFAR100, miniImageNet, and Caltech-USCD Birds-200-2011 (CUB200), show that our method significantly outperforms the baselines and sets new state-of-the-art results with remarkable advantages.
Few-shot class-incremental learning (FSCIL), which targets at continuously expanding model's representation capacity under few supervisions, is an important yet challenging problem. On the one hand, when fitting new tasks (novel classes), features trained on old tasks (old classes) could significantly drift, causing catastrophic forgetting. On the other hand, training the large amount of model parameters with few-shot novel-class examples leads to model over-fitting. In this paper, we propose a learnable expansion-and-compression network (LEC-Net), with the aim to simultaneously solve catastrophic forgetting and model over-fitting problems in a unified framework. By tentatively expanding network nodes, LEC-Net enlarges the representation capacity of features, alleviating feature drift of old network from the perspective of model regularization. By compressing the expanded network nodes, LEC-Net purses minimal increase of model parameters, alleviating over-fitting of the expanded network from a perspective of compact representation. Experiments on the CUB/CIFAR-100 datasets show that LEC-Net improves the baseline by 5~7% while outperforms the state-of-the-art by 5~6%. LEC-Net also demonstrates the potential to be a general incremental learning approach with dynamic model expansion capability.
In class-incremental learning, the objective is to learn a number of classes sequentially without having access to the whole training data. However, due to a problem known as catastrophic forgetting, neural networks suffer substantial performance drop in such settings. The problem is often approached by experience replay, a method which stores a limited number of samples to be replayed in future steps to reduce forgetting of the learned classes. When using a pretrained network as a feature extractor, we show that instead of training a single classifier incrementally, it is better to train a number of specialized classifiers which do not interfere with each other yet can cooperatively predict a single class. Our experiments on CIFAR100 dataset show that the proposed method improves the performance over SOTA by a large margin.
Despite their success for semantic segmentation, convolutional neural networks are ill-equipped for incremental learning, \ie, adapting the original segmentation model as new classes are available but the initial training data is not retained. Actually, they are vulnerable to catastrophic forgetting problem. We try to address this issue by "inverting" the trained segmentation network to synthesize input images starting from random noise. To avoid setting detailed pixel-wise segmentation maps as the supervision manually, we propose the SegInversion to synthesize images using the image-level labels. To increase the diversity of synthetic images, the Scale-Aware Aggregation module is integrated into SegInversion for controlling the scale (the number of pixels) of synthetic objects. Along with real images of new classes, the synthesized images will be fed into the distillation-based framework to train the new segmentation model which retains the information about previously learned classes, whilst updating the current model to learn the new ones. The proposed method significantly outperforms other incremental learning methods and obtains state-of-the-art performance on the PASCAL VOC 2012 and ADE20K datasets. The code and models will be made publicly available.
Deep learning models suffer from catastrophic forgetting when trained in an incremental learning setting. In this work, we propose a novel approach to address the task incremental learning problem, which involves training a model on new tasks that arrive in an incremental manner. The task incremental learning problem becomes even more challenging when the test set contains classes that are not part of the train set, i.e., a task incremental generalized zero-shot learning problem. Our approach can be used in both the zero-shot and non zero-shot task incremental learning settings. Our proposed method uses weight rectifications and affine transformations in order to adapt the model to different tasks that arrive sequentially. Specifically, we adapt the network weights to work for new tasks by "rectifying" the weights learned from the previous task. We learn these weight rectifications using very few parameters. We additionally learn affine transformations on the outputs generated by the network in order to better adapt them for the new task. We perform experiments on several datasets in both zero-shot and non zero-shot task incremental learning settings and empirically show that our approach achieves state-of-the-art results. Specifically, our approach outperforms the state-of-the-art non zero-shot task incremental learning method by over 5% on the CIFAR-100 dataset. Our approach also significantly outperforms the state-of-the-art task incremental generalized zero-shot learning method by absolute margins of 6.91% and 6.33% for the AWA1 and CUB datasets, respectively. We validate our approach using various ablation studies.
We address the problem of class incremental learning, which is a core step towards achieving adaptive vision intelligence. In particular, we consider the task setting of incremental learning with limited memory and aim to achieve better stability-plasticity trade-off. To this end, we propose a novel two-stage learning approach that utilizes a dynamically expandable representation for more effective incremental concept modeling. Specifically, at each incremental step, we freeze the previously learned representation and augment it with additional feature dimensions from a new learnable feature extractor. This enables us to integrate new visual concepts with retaining learned knowledge. We dynamically expand the representation according to the complexity of novel concepts by introducing a channel-level mask-based pruning strategy. Moreover, we introduce an auxiliary loss to encourage the model to learn diverse and discriminate features for novel concepts. We conduct extensive experiments on the three class incremental learning benchmarks and our method consistently outperforms other methods with a large margin.