A more realistic object detection paradigm, Open-World Object Detection, has arisen increasing research interests in the community recently. A qualified open-world object detector can not only identify objects of known categories, but also discover unknown objects, and incrementally learn to categorize them when their annotations progressively arrive. Previous works rely on independent modules to recognize unknown categories and perform incremental learning, respectively. In this paper, we provide a unified perspective: Semantic Topology. During the life-long learning of an open-world object detector, all object instances from the same category are assigned to their corresponding pre-defined node in the semantic topology, including the `unknown' category. This constraint builds up discriminative feature representations and consistent relationships among objects, thus enabling the detector to distinguish unknown objects out of the known categories, as well as making learned features of known objects undistorted when learning new categories incrementally. Extensive experiments demonstrate that semantic topology, either randomly-generated or derived from a well-trained language model, could outperform the current state-of-the-art open-world object detectors by a large margin, e.g., the absolute open-set error is reduced from 7832 to 2546, exhibiting the inherent superiority of semantic topology on open-world object detection.
Catastrophic forgetting in neural networks during incremental learning remains a challenging problem. Previous research investigated catastrophic forgetting in fully connected networks, with some earlier work exploring activation functions and learning algorithms. Applications of neural networks have been extended to include similarity learning. Understanding how similarity learning loss functions would be affected by catastrophic forgetting is of significant interest. Our research investigates catastrophic forgetting for four well-known similarity-based loss functions during incremental class learning. The loss functions are Angular, Contrastive, Center, and Triplet loss. Our results show that the catastrophic forgetting rate differs across loss functions on multiple datasets. The Angular loss was least affected, followed by Contrastive, Triplet loss, and Center loss with good mining techniques. We implemented three existing incremental learning techniques, iCaRL, EWC, and EBLL. We further proposed a novel technique using Variational Autoencoders (VAEs) to generate representation as exemplars passed through the network's intermediate layers. Our method outperformed three existing state-of-the-art techniques. We show that one does not require stored images (exemplars) for incremental learning with similarity learning. The generated representations from VAEs help preserve regions of the embedding space used by prior knowledge so that new knowledge does not ``overwrite'' it.
Loop closure detection is an important building block that ensures the accuracy and robustness of simultaneous localization and mapping (SLAM) systems. Due to their generalization ability, CNN-based approaches have received increasing attention. Although they normally benefit from training on datasets that are diverse and reflective of the environments, new environments often emerge after the model is deployed. It is therefore desirable to incorporate the data newly collected during operation for incremental learning. Nevertheless, simply finetuning the model on new data is infeasible since it may cause the model's performance on previously learned data to degrade over time, which is also known as the problem of catastrophic forgetting. In this paper, we present AirLoop, a method that leverages techniques from lifelong learning to minimize forgetting when training loop closure detection models incrementally. We experimentally demonstrate the effectiveness of AirLoop on TartanAir, Nordland, and RobotCar datasets. To the best of our knowledge, AirLoop is one of the first works to achieve lifelong learning of deep loop closure detectors.
Attack vectors are continuously evolving in order to evade Intrusion Detection systems. Internet of Things (IoT) environments, while beneficial for the IT ecosystem, suffer from inherent hardware limitations, which restrict their ability to implement comprehensive security measures and increase their exposure to vulnerability attacks. This paper proposes a novel Network Intrusion Prevention System that utilises a SelfOrganizing Incremental Neural Network along with a Support Vector Machine. Due to its structure, the proposed system provides a security solution that does not rely on signatures or rules and is capable to mitigate known and unknown attacks in real-time with high accuracy. Based on our experimental results with the NSL KDD dataset, the proposed framework can achieve on-line updated incremental learning, making it suitable for efficient and scalable industrial applications.
Unobtrusive and smart recognition of human activities using smartphones inertial sensors is an interesting topic in the field of artificial intelligence acquired tremendous popularity among researchers, especially in recent years. A considerable challenge that needs more attention is the real-time detection of physical activities, since for many real-world applications such as health monitoring and elderly care, it is required to recognize users' activities immediately to prevent severe damages to individuals' wellness. In this paper, we propose a human activity recognition (HAR) approach for the online prediction of physical movements, benefiting from the capabilities of incremental learning algorithms. We develop a HAR system containing monitoring software and a mobile application that collects accelerometer and gyroscope data and send them to a remote server via the Internet for classification and recognition operations. Six incremental learning algorithms are employed and evaluated in this work and compared with several batch learning algorithms commonly used for developing offline HAR systems. The Final results indicated that considering all performance evaluation metrics, Incremental K-Nearest Neighbors and Incremental Naive Bayesian outperformed other algorithms, exceeding a recognition accuracy of 95% in real-time.
In this paper, the issue of tailoring the soft confusion matrix (SCM) based classifier to deal with stream learning task is addressed. The main goal of the work is to develop a wrapping-classifier that allows incremental learning to classifiers that are unable to learn incrementally. The goal is achieved by making two improvements in the previously developed SCM classifier. The first one is aimed at reducing the computational cost of the SCM classifier. To do so, the definition of the fuzzy neighborhood of an object is changed. The second one is aimed at effective dealing with the concept drift. This is done by employing the ADWIN-driven concept drift detector that is not only used to detect the drift but also to control the size of the neighbourhood. The obtained experimental results show that the proposed approach significantly outperforms the reference methods.
Existing well investigated Predictive Process Monitoring techniques typically construct a predictive model based on past process executions, and then use it to predict the future of new ongoing cases, without the possibility of updating it with new cases when they complete their execution. This can make Predictive Process Monitoring too rigid to deal with the variability of processes working in real environments that continuously evolve and/or exhibit new variant behaviours over time. As a solution to this problem, we evaluate the use of three different strategies that allow the periodic rediscovery or incremental construction of the predictive model so as to exploit new available data. The evaluation focuses on the performance of the new learned predictive models, in terms of accuracy and time, against the original one, and uses a number of real and synthetic datasets with and without explicit Concept Drift. The results provide an evidence of the potential of incremental learning algorithms for predicting process monitoring in real environments.
Regularization-based methods are beneficial to alleviate the catastrophic forgetting problem in class-incremental learning. With the absence of old task images, they often assume that old knowledge is well preserved if the classifier produces similar output on new images. In this paper, we find that their effectiveness largely depends on the nature of old classes: they work well on classes that are easily distinguishable between each other but may fail on more fine-grained ones, e.g., boy and girl. In spirit, such methods project new data onto the feature space spanned by the weight vectors in the fully connected layer, corresponding to old classes. The resulting projections would be similar on fine-grained old classes, and as a consequence the new classifier will gradually lose the discriminative ability on these classes. To address this issue, we propose a memory-free generative replay strategy to preserve the fine-grained old classes characteristics by generating representative old images directly from the old classifier and combined with new data for new classifier training. To solve the homogenization problem of the generated samples, we also propose a diversity loss that maximizes Kullback Leibler (KL) divergence between generated samples. Our method is best complemented by prior regularization-based methods proved to be effective for easily distinguishable old classes. We validate the above design and insights on CUB-200-2011, Caltech-101, CIFAR-100 and Tiny ImageNet and show that our strategy outperforms existing memory-free methods with a clear margin. Code is available at https://github.com/xmengxin/MFGR
We describe federated reconnaissance, a class of learning problems in which distributed clients learn new concepts independently and communicate that knowledge efficiently. In particular, we propose an evaluation framework and methodological baseline for a system in which each client is expected to learn a growing set of classes and communicate knowledge of those classes efficiently with other clients, such that, after knowledge merging, the clients should be able to accurately discriminate between classes in the superset of classes observed by the set of clients. We compare a range of learning algorithms for this problem and find that prototypical networks are a strong approach in that they are robust to catastrophic forgetting while incorporating new information efficiently. Furthermore, we show that the online averaging of prototype vectors is effective for client model merging and requires only a small amount of communication overhead, memory, and update time per class with no gradient-based learning or hyperparameter tuning. Additionally, to put our results in context, we find that a simple, prototypical network with four convolutional layers significantly outperforms complex, state of the art continual learning algorithms, increasing the accuracy by over 22% after learning 600 Omniglot classes and over 33% after learning 20 mini-ImageNet classes incrementally. These results have important implications for federated reconnaissance and continual learning more generally by demonstrating that communicating feature vectors is an efficient, robust, and effective means for distributed, continual learning.
Ubiquitous personalized recommender systems are built to achieve two seemingly conflicting goals, to serve high quality content tailored to individual user's taste and to adapt quickly to the ever changing environment. The former requires a complex machine learning model that is trained on a large amount of data; the latter requires frequent update to the model. We present an incremental learning solution to provide both the training efficiency and the model quality. Our solution is based on sequential Bayesian update and quadratic approximation. Our focus is on large-scale personalized logistic regression models, with extensions to deep learning models. This paper fills in the gap between the theory and the practice by addressing a few implementation challenges that arise when applying incremental learning to large personalized recommender systems. Detailed offline and online experiments demonstrated our approach can significantly shorten the training time while maintaining the model accuracy. The solution is deployed in LinkedIn and directly applicable to industrial scale recommender systems.
Deep Learning (DL) algorithms are emerging as a key alternative to computationally expensive CFD simulations. However, state-of-the-art DL approaches require large and high-resolution training data to learn accurate models. The size and availability of such datasets are a major limitation for the development of next-generation data-driven surrogate models for turbulent flows. This paper introduces SURFNet, a transfer learning-based super-resolution flow network. SURFNet primarily trains the DL model on low-resolution datasets and transfer learns the model on a handful of high-resolution flow problems - accelerating the traditional numerical solver independent of the input size. We propose two approaches to transfer learning for the task of super-resolution, namely one-shot and incremental learning. Both approaches entail transfer learning on only one geometry to account for fine-grid flow fields requiring 15x less training data on high-resolution inputs compared to the tiny resolution (64x256) of the coarse model, significantly reducing the time for both data collection and training. We empirically evaluate SURFNet's performance by solving the Navier-Stokes equations in the turbulent regime on input resolutions up to 256x larger than the coarse model. On four test geometries and eight flow configurations unseen during training, we observe a consistent 2-2.1x speedup over the OpenFOAM physics solver independent of the test geometry and the resolution size (up to 2048x2048), demonstrating both resolution-invariance and generalization capabilities. Our approach addresses the challenge of reconstructing high-resolution solutions from coarse grid models trained using low-resolution inputs (super-resolution) without loss of accuracy and requiring limited computational resources.
Lyrics play a significant role in conveying the song's mood and are information to understand and interpret music communication. Conventional natural language processing approaches use translation of the Hindi text into English for analysis. This approach is not suitable for lyrics as it is likely to lose the inherent intended contextual meaning. Thus, the need was identified to develop a system for Devanagari text analysis. The data set of 300 song lyrics with equal distribution in five different moods is used for the experimentation. The proposed system performs contextual mood analysis of Hindi song lyrics in Devanagari text format. The contextual analysis is stored as a knowledge base, updated using an incremental learning approach with new data. Contextual knowledge graph with moods and associated important contextual terms provides the graphical representation of the lyric data set used. The testing results show 64% accuracy for the mood prediction. This work can be easily extended to applications related to Hindi literary work such as summarization, indexing, contextual retrieval, context-based classification and grouping of documents.
To accommodate rapid changes in the real world, the cognition system of humans is capable of continually learning concepts. On the contrary, conventional deep learning models lack this capability of preserving previously learned knowledge. When a neural network is fine-tuned to learn new tasks, its performance on previously trained tasks will significantly deteriorate. Many recent works on incremental object detection tackle this problem by introducing advanced regularization. Although these methods have shown promising results, the benefits are often short-lived after the first incremental step. Under multi-step incremental learning, the trade-off between old knowledge preserving and new task learning becomes progressively more severe. Thus, the performance of regularization-based incremental object detectors gradually decays for subsequent learning steps. In this paper, we aim to alleviate this performance decay on multi-step incremental detection tasks by proposing a dilatable incremental object detector (DIODE). For the task-shared parameters, our method adaptively penalizes the changes of important weights for previous tasks. At the same time, the structure of the model is dilated or expanded by a limited number of task-specific parameters to promote new task learning. Extensive experiments on PASCAL VOC and COCO datasets demonstrate substantial improvements over the state-of-the-art methods. Notably, compared with the state-of-the-art methods, our method achieves up to 6.0% performance improvement by increasing the number of parameters by just 1.2% for each newly learned task.
Face forgery by deepfake is widely spread over the internet and this raises severe societal concerns. In this paper, we propose a novel video transformer with incremental learning for detecting deepfake videos. To better align the input face images, we use a 3D face reconstruction method to generate UV texture from a single input face image. The aligned face image can also provide pose, eyes blink and mouth movement information that cannot be perceived in the UV texture image, so we use both face images and their UV texture maps to extract the image features. We present an incremental learning strategy to fine-tune the proposed model on a smaller amount of data and achieve better deepfake detection performance. The comprehensive experiments on various public deepfake datasets demonstrate that the proposed video transformer model with incremental learning achieves state-of-the-art performance in the deepfake video detection task with enhanced feature learning from the sequenced data.
Incremental learning of semantic segmentation has emerged as a promising strategy for visual scene interpretation in the open- world setting. However, it remains challenging to acquire novel classes in an online fashion for the segmentation task, mainly due to its continuously-evolving semantic label space, partial pixelwise ground-truth annotations, and constrained data availability. To ad- dress this, we propose an incremental learning strategy that can fast adapt deep segmentation models without catastrophic forgetting, using a streaming input data with pixel annotations on the novel classes only. To this end, we develop a uni ed learning strategy based on the Expectation-Maximization (EM) framework, which integrates an iterative relabeling strategy that lls in the missing labels and a rehearsal-based incremental learning step that balances the stability-plasticity of the model. Moreover, our EM algorithm adopts an adaptive sampling method to select informative train- ing data and a class-balancing training strategy in the incremental model updates, both improving the e cacy of model learning. We validate our approach on the PASCAL VOC 2012 and ADE20K datasets, and the results demonstrate its superior performance over the existing incremental methods.
Catastrophic forgetting describes the fact that machine learning models will likely forget the knowledge of previously learned tasks after the learning process of a new one. It is a vital problem in the continual learning scenario and recently has attracted tremendous concern across different communities. In this paper, we explore the catastrophic forgetting phenomena in the context of quantum machine learning. We find that, similar to those classical learning models based on neural networks, quantum learning systems likewise suffer from such forgetting problem in classification tasks emerging from various application scenes. We show that based on the local geometrical information in the loss function landscape of the trained model, a uniform strategy can be adapted to overcome the forgetting problem in the incremental learning setting. Our results uncover the catastrophic forgetting phenomena in quantum machine learning and offer a practical method to overcome this problem, which opens a new avenue for exploring potential quantum advantages towards continual learning.
This paper presents a novel method for attitude estimation of an object in 3D space by incremental learning of the Long-Short Term Memory (LSTM) network. Gyroscope, accelerometer, and magnetometer are few widely used sensors in attitude estimation applications. Traditionally, multi-sensor fusion methods such as the Extended Kalman Filter and Complementary Filter are employed to fuse the measurements from these sensors. However, these methods exhibit limitations in accounting for the uncertainty, unpredictability, and dynamic nature of the motion in real-world situations. In this paper, the inertial sensors data are fed to the LSTM network which are then updated incrementally to incorporate the dynamic changes in motion occurring in the run time. The robustness and efficiency of the proposed framework is demonstrated on the dataset collected from a commercially available inertial measurement unit. The proposed framework offers a significant improvement in the results compared to the traditional method, even in the case of a highly dynamic environment. The LSTM framework-based attitude estimation approach can be deployed on a standard AI-supported processing module for real-time applications.
Traditional learning systems are trained in closed-world for a fixed number of classes, and need pre-collected datasets in advance. However, new classes often emerge in real-world applications and should be learned incrementally. For example, in electronic commerce, new types of products appear daily, and in a social media community, new topics emerge frequently. Under such circumstances, incremental models should learn several new classes at a time without forgetting. We find a strong correlation between old and new classes in incremental learning, which can be applied to relate and facilitate different learning stages mutually. As a result, we propose CO-transport for class Incremental Learning (COIL), which learns to relate across incremental tasks with the class-wise semantic relationship. In detail, co-transport has two aspects: prospective transport tries to augment the old classifier with optimal transported knowledge as fast model adaptation. Retrospective transport aims to transport new class classifiers backward as old ones to overcome forgetting. With these transports, COIL efficiently adapts to new tasks, and stably resists forgetting. Experiments on benchmark and real-world multimedia datasets validate the effectiveness of our proposed method.
Although the concept of catastrophic forgetting is straightforward, there is a lack of study on its causes. In this paper, we systematically explore and reveal three causes for catastrophic forgetting in Class Incremental Learning(CIL). From the perspective of representation learning,(i) intra-phase forgetting happens when the learner fails to correctly align the same-phase data as training proceeds and (ii) inter-phase confusion happens when the learner confuses the current-phase data with the previous-phase. From the task-specific point of view, the CIL model suffers from the problem of (iii) classifier deviation. After investigating existing strategies, we observe that there is a lack of study on how to prevent the inter-phase confusion. To initiate the research on this specific issue, we propose a simple yet effective framework, Contrastive Class Concentration for CIL (C4IL). Our framework leverages the class concentration effect of contrastive learning, yielding a representation distribution with better intra-class compactibility and inter-class separability. Empirically, we observe that C4IL significantly lowers the probability of inter-phase confusion and as a result improves the performance on multiple CIL settings of multiple datasets.
Catastrophic forgetting is one of the major challenges on the road for continual learning systems, which are presented with an on-line stream of tasks. The field has attracted considerable interest and a diverse set of methods have been presented for overcoming this challenge. Learning without Forgetting (LwF) is one of the earliest and most frequently cited methods. It has the advantages of not requiring the storage of samples from the previous tasks, of implementation simplicity, and of being well-grounded by relying on knowledge distillation. However, the prevailing view is that while it shows a relatively small amount of forgetting when only two tasks are introduced, it fails to scale to long sequences of tasks. This paper challenges this view, by showing that using the right architecture along with a standard set of augmentations, the results obtained by LwF surpass the latest algorithms for task incremental scenario. This improved performance is demonstrated by an extensive set of experiments over CIFAR-100 and Tiny-ImageNet, where it is also shown that other methods cannot benefit as much from similar improvements.