Lifelong Learning Scenarios Research Articles

Advancing Neuro-Inspired Lifelong Learning for Edge with Co-Design

Lifelong learning, which refers to an agent's ability to continuously learn and enhance its performance over its lifespan, is a significant challenge in artificial intelligence (AI), that biological systems tackle efficiently. This challenge is further exacerbated when AI is deployed in untethered environments with strict energy and latency constraints. We take inspiration from neural plasticity and investigate how to leverage and build energy-efficient lifelong learning machines. Specifically, we study how a combination of neural plasticity mechanisms, namely neuromodulation, synaptic consolidation, and metaplasticity, enhance the continual learning capabilities of AI models. We further co-design architectures that leverage compute-in-memory topologies and sparse spike-based communication with quantization for the edge. Aspects of this co-design can be transferred to federated lifelong learning scenarios.

Sparsify dynamically expandable network via variational dropout

This paper develops a new method for lifelong learning referred to as Sparsify Dynamically Expandable Network (SDEN) via Variational Dropout, which explores a sparse model while preserving the performance. Dynamically Expandable Network (DEN) can learn a sequence of tasks via performing network retraining, network expansion by adding only the necessary neurons, and network split to effectively prevent semantic drift in an online manner. To overcome point estimation of parameters in DEN, Bayesian Compression for DEN is developed under the Bayesian framework. However, this method demands more time for model training and testing. To improve the model efficiency, we propose SDEN under the efficient sparse learning framework. We validate our SDEN in the lifelong learning scenarios with multiple frequently used benchmarks, on which it can obtain comparable classification accuracy, and less training and testing time compared with the comparison methods. Furthermore, our method can also learn a more sparse network structure which means fewer network parameters.

Lifelong learning with Shared and Private Latent Representations learned through synaptic intelligence

This paper explores a novel lifelong learning method with Shared and Private Latent Representations (SPLR), which are learned through synaptic intelligence. To solve a sequence of tasks, by considering the entire parameter learning trajectory, SPLR can learn task-invariant representation which changes little, and task-specific features that change greatly along the entire parameter updating trajectory. Therefore, in the lifelong learning scenarios, our model can obtain a task-invariant structure shared by all tasks and also contain some private properties that are task-specific to each task. To reduce the parameter quantity, a ℓ1 regularization to promote sparsity is employed in the weights. We use multiple datasets under lifelong learning scenes to verify our SPLR, on these datasets it can get comparable performance compared with existing lifelong learning approaches, and learn a sparse network which means fewer parameters while requiring less model training time.

Study of a lifelong learning scenario for question answering

Question Answering (QA) systems have witnessed a significant advance in the last years due to the development of neural architectures employing pre-trained large models like BERT. However, once the QA model is fine-tuned for a task (e.g., a particular type of questions over a particular domain), system performance drops when new tasks are added along time, (e.g., new types of questions or new domains). Therefore, the system requires a retraining but, since the data distribution has shifted away from the previous learning, performance over previous tasks drops significantly. Hence, we need strategies to make our systems resistant to the passage of time. Lifelong Learning (LL) aims to study how systems can take advantage of the previous learning and the knowledge acquired to maintain or improve performance over time. In this article, we explore a scenario where the same LL based QA system suffers along time several shifts in the data distribution, represented as the addition of new different QA datasets. In this setup, the following research questions arise: (i) How LL based QA systems can benefit from previously learned tasks? (ii) Is there any strategy general enough to maintain or improve the performance over time when new tasks are added? and finally, (iii) How to detect a lack of knowledge that impedes the answering of questions and must trigger a new learning process? To answer these questions, we systematically try all possible training sequences over three well known QA datasets. Our results show how the learning of a new dataset is sensitive to previous training sequences and that we can find a strategy general enough to avoid the combinatorial explosion of testing all possible training sequences. Thus, when a new dataset is added to the system, the best way to retrain the system without dropping performance over the previous datasets is to randomly merge the new training material with the previous one.

Chunk Dynamic Updating for Group Lasso with ODEs

Group Lasso is an important sparse regression method in machine learning which encourages selecting key explanatory factors in a grouped manner because of the use of L-2,1 norm. In real-world learning tasks, some chunks of data would be added into or removed from the training set in sequence due to the existence of new or obsolete historical data, which is normally called dynamic or lifelong learning scenario. However, most of existing algorithms of group Lasso are limited to offline updating, and only one is online algorithm which can only handle newly added samples inexactly. Due to the complexity of L-2,1 norm, how to achieve accurate chunk incremental and decremental learning efficiently for group Lasso is still an open question. To address this challenging problem, in this paper, we propose a novel accurate dynamic updating algorithm for group Lasso by utilizing the technique of Ordinary Differential Equations (ODEs), which can incorporate or eliminate a chunk of samples from original training set without retraining the model from scratch. Specifically, we introduce a new formulation to reparameterize the adjustment procedures of chunk incremental and decremental learning simultaneously. Based on the new formulation, we propose a path following algorithm for group Lasso regarding to the adjustment parameter. Importantly, we prove that our path following algorithm can exactly track the piecewise smooth solutions thanks to the technique of ODEs, so that the accurate chunk incremental and decremental learning can be achieved. Extensive experimental results not only confirm the effectiveness of proposed algorithm for the chunk incremental and decremental learning, but also validate its efficiency compared to the existing offline and online algorithms.

Cross-modal self-supervised representation learning for gesture and skill recognition in robotic surgery.

Multi- and cross-modal learning consolidates information from multiple data sources which may offer a holistic representation of complex scenarios. Cross-modal learning is particularly interesting, because synchronized data streams are immediately useful as self-supervisory signals. The prospect of achieving self-supervised continual learning in surgical robotics is exciting as it may enable lifelong learning that adapts to different surgeons and cases, ultimately leading to a more general machine understanding of surgical processes. We present a learning paradigm using synchronous video and kinematics from robot-mediated surgery. Our approach relies on an encoder-decoder network that maps optical flow to the corresponding kinematics sequence. Clustering on the latent representations reveals meaningful groupings for surgeon gesture and skill level. We demonstrate the generalizability of the representations on the JIGSAWS dataset by classifying skill and gestures on tasks not used for training. For tasks seen in training, we report a 59 to 70% accuracy in surgical gestures classification. On tasks beyond the training setup, we note a 45 to 65% accuracy. Qualitatively, we find that unseen gestures form clusters in the latent space of novice actions, which may enable the automatic identification of novel interactions in a lifelong learning scenario. From predicting the synchronous kinematics sequence, optical flow representations of surgical scenes emerge that separate well even for new tasks that the model had not seen before. While the representations are useful immediately for a variety of tasks, the self-supervised learning paradigm may enable research in lifelong and user-specific learning.

Continual Multiview Task Learning via Deep Matrix Factorization.

The state-of-the-art multitask multiview (MTMV) learning tackles a scenario where multiple tasks are related to each other via multiple shared feature views. However, in many real-world scenarios where a sequence of the multiview task comes, the higher storage requirement and computational cost of retraining previous tasks with MTMV models have presented a formidable challenge for this lifelong learning scenario. To address this challenge, in this article, we propose a new continual multiview task learning model that integrates deep matrix factorization and sparse subspace learning in a unified framework, which is termed deep continual multiview task learning (DCMvTL). More specifically, as a new multiview task arrives, DCMvTL first adopts a deep matrix factorization technique to capture hidden and hierarchical representations for this new coming multiview task while accumulating the fresh multiview knowledge in a layerwise manner. Then, a sparse subspace learning model is employed for the extracted factors at each layer and further reveals cross-view correlations via a self-expressive constraint. For model optimization, we derive a general multiview learning formulation when a new multiview task comes and apply an alternating minimization strategy to achieve lifelong learning. Extensive experiments on benchmark data sets demonstrate the effectiveness of our proposed DCMvTL model compared with the existing state-of-the-art MTMV and lifelong multiview task learning models.

Lifelong Learning of Spatiotemporal Representations With Dual-Memory Recurrent Self-Organization.

Artificial autonomous agents and robots interacting in complex environments are required to continually acquire and fine-tune knowledge over sustained periods of time. The ability to learn from continuous streams of information is referred to as lifelong learning and represents a long-standing challenge for neural network models due to catastrophic forgetting in which novel sensory experience interferes with existing representations and leads to abrupt decreases in the performance on previously acquired knowledge. Computational models of lifelong learning typically alleviate catastrophic forgetting in experimental scenarios with given datasets of static images and limited complexity, thereby differing significantly from the conditions artificial agents are exposed to. In more natural settings, sequential information may become progressively available over time and access to previous experience may be restricted. Therefore, specialized neural network mechanisms are required that adapt to novel sequential experience while preventing disruptive interference with existing representations. In this paper, we propose a dual-memory self-organizing architecture for lifelong learning scenarios. The architecture comprises two growing recurrent networks with the complementary tasks of learning object instances (episodic memory) and categories (semantic memory). Both growing networks can expand in response to novel sensory experience: the episodic memory learns fine-grained spatiotemporal representations of object instances in an unsupervised fashion while the semantic memory uses task-relevant signals to regulate structural plasticity levels and develop more compact representations from episodic experience. For the consolidation of knowledge in the absence of external sensory input, the episodic memory periodically replays trajectories of neural reactivations. We evaluate the proposed model on the CORe50 benchmark dataset for continuous object recognition, showing that we significantly outperform current methods of lifelong learning in three different incremental learning scenarios.

Developing Student Support for Open and Distance Learning: The EMPOWER Project

European universities face great challenges dealing with twenty-first century world changes. Deep transformations are required to a wide range of life-long learning scenarios, which are replacing traditional modes of university study and giving access to students in more flexible ways. To address the transformation in learning, the European Association of Distance Teaching Universities (EADTU) launched the EMPOWER project in 2015 so that the EADTU could share expertise of distance education universities in the field.The EMPOWER project is organised in 12 areas. One is student support, as the central area for students’ success. The project’s goal is to empower students to become life-long, self-directed learners in open, online and blended-learning environments. The plan was to increase student retention and enhance academic performance, integration and satisfaction.This article summarises the work that has been done during the past two years to offer different tools and resources, such as webinars and reports. These can help institutions and academics in their understanding of what underlies student engagement and motivation versus student drop-out. We aim to share expertise about how effective, or not, technology has been for developing innovative, advanced and quality student support services to large and small groups of students.It is too early to judge the success of the student support group of the EMPOWER project. This paper looks at what has been achieved so far. In particular it outlines the reasons for focusing on the problem of student drop-out and how student support may help to ameliorate the problem.

Interaction for lifelong learning

AbstractThe term ‘lifelong’, as applied to education or learning, has been in circulation for more than a quarter of a century. It has played an important role in policy discussions, as well as in studies of the sociology and economics of education. The relationship of this term to the rapidly changing world of information and educational technologies, and to the various conceptions of interaction that are central to these technologies, however, has been considered much less frequently. This paper seeks to shed light on the relationship between lifelong learning and the interactive technologies that have become associated specifically with the Semantic Web. It begins by presenting a fictional narrative to illustrate a lifelong learning scenario in the context of the services and resources that the Semantic Web will be capable of providing. It then proceeds to isolate a number of general characteristics of lifelong learning as they are manifest in this scenario and in recent literature on the subject. The paper then explores how emergent, interactive technologies of the Semantic Web have the general potential to address many of the characteristics of lifelong learning, and hold out the promise of satisfying a wide variety of lifelong learning needs. It will conclude by considering some of the outstanding challenges presented by lifelong learning contexts, and mention some of the limitations of advanced technologies used to address these needs.