Previous Task Research Articles

Class-incremental object detection

Deep learning architectures have shown remarkable results in the object detection task. However, they experience a critical performance drop when they are required to learn new classes incrementally without forgetting old ones. This catastrophic forgetting phenomenon impedes the deployment of artificial intelligence in real word scenarios where systems need to learn new and different representations over time. Recently, many incremental learning methods have been proposed to avoid the catastrophic forgetting problem. However, current state-of-the-art class-incremental learning strategies aim at preserving the knowledge of old classes while learning new ones sequentially, which would encounter other problems as follows: (1) In the process of preserving information of old classes, only a small portion of data in the previous tasks are kept and replayed during training, which inevitably incurs bias that is favorable for the new classes but malicious to the old classes. (2) With the knowledge of previous classes distilled into the new model, a sub-optimal solution for the new task is obtained since the preserving process of previous classes sabotages the training of new classes. To address these issues, termed as Information Asymmetry (IA), we propose a double-head framework which preserves the knowledge of old classes and learns the knowledge of new classes separately. Specifically, we transfer the knowledge of the previous model to the current learned one for overcoming the catastrophic forgetting problem. Furthermore, considering that IA would introduce impacts on the training of the new model, we propose a Non-Affection mask to distill the knowledge of the interested regions at the feature level. Comprehensive experimental results demonstrate that our proposed method significantly outperforms other state-of-the-art class-incremental object detection methods on PASCAL VOC and MS COCO datasets.

The Blinded-Dose Purchase Task: assessing hypothetical demand based on cocaine, methamphetamine, and alcohol administration

RationaleBehavioral economic drug purchase tasks quantify the reinforcing value of a drug (i.e., demand). Although widely used to assess demand, drug expectancies are rarely accounted for and may introduce variability across participants given diverse drug experiences.ObjectivesThree experiments validated and extended previous hypothetical purchase tasks by using blinded drug dose as a reinforcing stimulus, and determined hypothetical demand for experienced effects while controlling for drug expectancies.MethodsAcross three double-blind, placebo-controlled, within-subject experiments, cocaine (0, 125, 250 mg/70 kg; n=12), methamphetamine (0, 20, 40 mg; n=19), and alcohol (0, 1 g/kg alcohol; n=25) were administered and demand was assessed using the Blinded-Dose Purchase Task. Participants answered questions regarding simulated purchasing of the blinded drug dose across increasing prices. Demand metrics, subjective effects, and self-reported real-world monetary spending on drugs were evaluated.ResultsData were well modeled by the demand curve function, with significantly higher intensity (purchasing at low prices) for active drug doses compared to placebo for all experiments. Unit-price analyses revealed more persistent consumption across prices (lower α) in the higher compared to lower active dose condition for methamphetamine (a similar non-significant finding emerged for cocaine). Significant associations between demand metrics, peak subjective effects, and real-world spending on drugs also emerged across all experiments.ConclusionsOrderly demand curve data revealed differences across drug and placebo conditions, and relations to real-world measures of drug spending, and subjective effects. Unit-price analyses enabled parsimonious comparisons across doses. Results lend credence to the validity of the Blinded-Dose Purchase Task, which allows for control of drug expectancies.

Assign and Appraise: Achieving Optimal Performance in Collaborative Teams

Tackling complex team problems requires understanding each team member’s skills in order to devise a task assignment maximizing the team performance. This article proposes a novel quantitative model describing the decentralized process by which individuals in a team learn who has what abilities, while concurrently assigning tasks to each of the team members. In the model, the appraisal network represents team members’ evaluations of one another, and each team member chooses their own workload. The appraisals and workload assignment change simultaneously: each member builds their own local appraisal of neighboring members based on the performance exhibited on previous tasks, while the workload is redistributed based on the current appraisal estimates. We show that the appraisal states can be reduced to a lower dimension due to the presence of conserved quantities associated with the cycles of the appraisal network. Building on this, we provide rigorous results characterizing the ability, or inability, of the team to learn each other’s skills and, thus, converge to an allocation maximizing the team performance. We complement our analysis with extensive numerical experiments.

Lifelong Online Learning from Accumulated Knowledge

In this article, we formulate lifelong learning as an online transfer learning procedure over consecutive tasks, where learning a given task depends on the accumulated knowledge. We propose a novel theoretical principled framework, lifelong online learning, where the learning process for each task is in an incremental manner. Specifically, our framework is composed of two-level predictions: the prediction information that is solely from the current task; and the prediction from the knowledge base by previous tasks. Moreover, this article tackled several fundamental challenges: arbitrary or even non-stationary task generation process, an unknown number of instances in each task, and constructing an efficient accumulated knowledge base. Notably, we provide a provable bound of the proposed algorithm, which offers insights on the how the accumulated knowledge improves the predictions. Finally, empirical evaluations on both synthetic and real datasets validate the effectiveness of the proposed algorithm.

To Do or Not to Do: The cerebellum and neocortex contribute to predicting sequences of social intentions

Humans read the minds of others to predict their actions and efficiently navigate social environments, a capacity called mentalizing. Accumulating evidence suggests that the cerebellum, especially Crus 1 and 2, and lobule IX are involved in identifying the sequence of others’ actions. In the current study, we investigated the neural correlates that underly predicting others’ intentions and how this plays out in the sequence of their actions. We developed a novel intention prediction task, which required participants to put protagonists’ behaviors in the correct chronological order based on the protagonists’ honest or deceitful intentions (i.e., inducing true or false beliefs in others). We found robust activation of cerebellar lobule IX and key mentalizing areas in the neocortex when participants ordered protagonists’ intentional behaviors compared with not ordering behaviors or to ordering object scenarios. Unlike a previous task that involved prediction based on personality traits that recruited cerebellar Crus 1 and 2, and lobule IX (Haihambo et al., 2021), the present task recruited only the cerebellar lobule IX. These results suggest that cerebellar lobule IX may be generally involved in social action sequence prediction, and that different areas of the cerebellum are specialized for distinct mentalizing functions.

Federated probability memory recall for federated continual learning

Federated Continual Learning (FCL) approaches exist two major problems of the probability bias and the imbalance in parameter variations. These two problems lead to catastrophic forgetting of the network in the FCL process. Therefore, this paper proposes a novel FCL framework, Federated Probability Memory Recall (FedPMR), to mitigate the probability bias problem and the imbalance in parameter variations. Firstly, for the probability bias problem, this paper designs the Probability Distribution Alignment (PDA) module, which consolidates the memory of old probability experience. Specifically, PDA maintains a replay buffer and uses the probability memory stored in the buffer to correct the offset probabilities of the previous tasks during the two-stage training. Secondly, to alleviate the imbalance in parameter variations, this paper designs the Parameter Consistency Constraint (PCC) module, which constrains the magnitude of neural weight changes for previous tasks. Concretely, PCC applies a set of adaptive weights to subsets of the regularization term that constrains parameter changes, forcing the current model to be sufficiently close to the past model in parameter space distance. Experiments with various levels of task similitude across clients demonstrate that our technique establishes the new state-of-the-art performance when compared to previous FCL approaches.

Improving transparency and representational generalizability through parallel continual learning

This paper takes a parallel learning approach in continual learning scenarios. We define parallel continual learning as learning a sequence of tasks where the data for the previous tasks, whose distribution may have shifted over time, are also available while learning new tasks. We propose a parallel continual learning method by assigning subnetworks to each task, and simultaneously training only the assigned subnetworks on their corresponding tasks. In doing so, some parts of the network will be shared across multiple tasks. This is unlike the existing literature in continual learning which aims at learning incoming tasks sequentially, with the assumption that the data for the previous tasks have a fixed distribution. Our proposed method offers promises in: (1) Transparency in the network and in the relationship across tasks by enabling examination of the learned representations by independent and shared subnetworks, (2) Representation generalizability through sharing and training subnetworks on multiple tasks simultaneously. Our analysis shows that compared to many competing approaches such as continual learning, neural architecture search, and multi-task learning, parallel continual learning is capable of learning more generalizable representations. Also, (3)Parallel continual learning overcomes the common issue of catastrophic forgetting in continual learning algorithms. This is the first effort to train a neural network on multiple tasks and input domains simultaneously in a continual learning scenario. Our code is available at https://github.com/yours-anonym/PaRT.

CLELNet: A continual learning network for esophageal lesion analysis on endoscopic images

Background and Objective: A deep learning-based intelligent diagnosis system can significantly reduce the burden of endoscopists in the daily analysis of esophageal lesions. Considering the need to add new tasks in the diagnosis system, a deep learning model that can train a series of tasks incrementally using endoscopic images is essential for identifying the types and regions of esophageal lesions.Method: In this paper, we proposed a continual learning-based esophageal lesion network (CLELNet), in which a convolutional autoencoder was designed to extract representation features of endoscopic images among different esophageal lesions. The proposed CLELNet consists of shared layers and task-specific layers. Shared layers are used to extract common features among different lesions while task-specific layers can complete different tasks. The first two tasks trained by the CLELNet are the classification (task 1) and the segmentation (task 2). We collected a dataset of esophageal endoscopic images from Macau Kiang Wu Hospital for training and testing the CLELNet.Results: The experimental results showed that the classification accuracy of task 1 was 95.96%, and the Intersection Over Union and the Dice Similarity Coefficient of task 2 were 65.66% and 78.08%, respectively.Conclusions: The proposed CLELNet can realize task-incremental learning without forgetting the previous tasks and thus become a useful computer-aided diagnosis system in esophageal lesions analysis.

Towards well-generalizing meta-learning via adversarial task augmentation

Meta-learning aims to use the knowledge from previous tasks to facilitate the learning of novel tasks. Many meta-learning models elaborately design various task-shared inductive bias, and learn it from a large number of tasks, so the generalization capability of the learned inductive bias depends on the diversity of the training tasks. A common assumption in meta-learning is that the training tasks and the test tasks come from the same or similar task distributions. However, this is usually not strictly satisfied in practice, so meta-learning models need to cope with various novel in-domain or cross-domain tasks. To this end, we propose to use task augmentation to increase the diversity of training tasks, thereby improving the generalization capability of meta-learning models. Concretely, we consider the worst-case problem around the base task distribution, and derive the adversarial task augmentation method which can generate inductive bias-adaptive ‘challenging’ tasks. Our method can be used as a simple plug-and-play module for various meta-learning models, and improve their generalization capability. We conduct extensive experiments under in-domain and cross-domain few-shot learning and unsupervised few-shot learning settings, and evaluate our method on different types of data (images and text). Experimental results show that our method can effectively improve the generalization capability of various meta-learning models under different settings.

Continual relation extraction via linear mode connectivity and interval cross training

Continual relation extraction techniques aim to meet the requirements of real-world applications, in which new data and relations emerge constantly. To strike a balance between continual learning ability and computational cost, many previous works on replay-based methods emphasized on employing different strategies to use a smaller number of stored replay samples to maintain the continual ability of the model. However, catastrophic forgetting is prone to occur if a limited number of previously stored replay samples have been used. In this paper, we proposed a continual learning method which contains three learning phases, namely preliminary learning, memory retention, and memory reconsolidation. In more detail, the model is fine-tuned on the model of the last previous task firstly in the preliminary learning stage. During the memory retention stage, the method utilizes the characteristics of the model’s linear mode connectivity between the multi-task learning model and the continual learning model to retain the knowledge of previous tasks as much as possible. At the memory reconsolidation stage, it uses the previously stored samples as well as the ones of the current task for interval cross training to reconsolidate the learned knowledge. In order to evaluate the continual learning ability of the model learning by the proposed method, we tested it on three well-studied benchmarks, and the experimental results show that our algorithm performs better than the other compared algorithms in terms of the average accuracy as well as overall accuracy.

A transfer learning framework for well placement optimization based on denoising autoencoder

Well placement optimization is directly related to the recovery factor of reservoir development, and at present, the mainstream solution is an evolutionary algorithm. However, time-consuming numerical simulators need to be called to evaluate each alternative well placement scheme. Since the rules of well placement problems are universal, similar reservoirs will have similar well locations. Thus, knowledge transfer across similar well placement optimization tasks can expedite searching effectively. To this end, this paper proposes a novel transfer learning framework for well placement optimization to extract the potential well placement rules based on the feature extraction capability of a single-layer denoising autoencoder. The reconstruction mapping between the previous and present tasks is established to make the randomly generated well locations inherit the knowledge from the optimal well locations of the previous task, which helps the search direction of the evolutionary algorithm quickly bias to the optimal solution, thus, the solving of present task can be accelerated. The simplified denoising autoencoder holds a closed-form solution after derivation of the loss function, and the corresponding reuse of knowledge will not bring much additional computational burden on the evolutionary search. In addition, a similarity measure method between well placement optimization tasks is proposed to avoid a negative transfer. At last, comprehensive experiments on benchmark functions and well placement optimization instances are presented to evaluate the effectiveness of the proposed framework.

A Knowledge-Grounded Task-Oriented Dialogue System with Hierarchical Structure for Enhancing Knowledge Selection

For a task-oriented dialogue system to provide appropriate answers to and services for users' questions, it is necessary for it to be able to utilize knowledge related to the topic of the conversation. Therefore, the system should be able to select the most appropriate knowledge snippet from the knowledge base, where external unstructured knowledge is used to respond to user requests that cannot be solved by the internal knowledge addressed by the database or application programming interface. Therefore, this paper constructs a three-step knowledge-grounded task-oriented dialogue system with knowledge-seeking-turn detection, knowledge selection, and knowledge-grounded generation. In particular, we propose a hierarchical structure of domain-classification, entity-extraction, and snippet-ranking tasks by subdividing the knowledge selection step. Each task is performed through the pre-trained language model with advanced techniques to finally determine the knowledge snippet to be used to generate a response. Furthermore, the domain and entity information obtained because of the previous task is used as knowledge to reduce the search range of candidates, thereby improving the performance and efficiency of knowledge selection and proving it through experiments.

Non-Conscious Negativity Bias as a Predictor of Depression Severity in TMS Patients

Abstract Background: Patients who undergo TMS (transcranial magnetic stimulation) are typically assessed using subjective rating scales. However, more objective assessments may lead to the development of biomarkers that may be better able to predict outcomes after TMS. In this study, we evaluated the non-conscious negativity bias test as an objective measure of depression severity. Design: This was a retrospective study of a convenience sample of 8 patients undergoing TMS treatment for major depression. Patients took a test of non-conscious negativity bias from a computerized platform (Total Brain). This test involved patients recognizing a facial emotional expression (fear, anger, disgust, sadness, happiness, or neutrality) as quickly as possible. After this first task, a set of two faces was presented, with one face repeated from the previous first task, and one face new. Both faces displayed the same emotion, and the task was to select the face seen from the previous task. Raw scores were time to correct responses for each of the four negative emotions (fear, sadness, anger, and disgust). Raw scores were converted to a percentile score based on age, education level, and gender. Percentiles of 1-30 were categorized as low, 30-70 as medium, and 70-100 as high. Results: Out of 4 patients who scored in the low category on the test, all of them (100%) rated their depression on the PHQ9 in the moderately severe to severe range (15-27). Out of the 4 who were not in the low category on the test, 1 rated their depression on the PHQ9 as severe (25%). Using the Fisher Exact test, the association between the variables approached significance (p=0.14). Discussion: In this retrospective study, results from an implicit negative bias test were suggestive of depression severity in patients undergoing TMS. A larger sample size may help clarify its use in real-world populations. Research Category and Technology and Methods Clinical Research: 10. Transcranial Magnetic Stimulation (TMS) Keywords: TMS, Depression, Biomarker, Cognitive

Incremental Pedestrian Attribute Recognition via Dual Uncertainty-Aware Pseudo-Labeling

Incremental pedestrian attribute recognition (IncPAR) aims to learn novel person attributes continuously and avoid the catastrophic forgetting, which is an essential problem for image forensic and security applications, e.g., suspect search. Different from the conventional continual learning for visual classification, we formulate the IncPAR as a problem of multi-label continual learning with incomplete labels (MCL-IL), where the training samples in a novel task are annotated with only a few categories of interest but may implicitly contain other attributes of previous tasks. The incomplete label assignments is a challenging and frequently-encountered issue in real-world multi-label classification applications due to a number of reasons, e.g., incomplete data collection, moderate budget for annotations, etc. To tackle the MCL-IL problem, we propose a self-training based approach via dual uncertainty-aware pseudo-labeling (DUAPL) to transfer the knowledge learned in previous tasks to novel tasks. Specially, both kinds of uncertainties, i.e., aleatoric uncertainty and epistemic uncertainty, are modeled to mitigate the negative influences of noisy pseudo labels induced by low quality samples and immature models learned by inadequate training in early tasks. Based on the DUAPL, more reliable supervision signals can be estimated to prevent the model evolution from forgetting attributes seen in previous tasks. For standard evaluations of MCL-IL methods, two benchmarks on IncPAR, termed RAP-CL and PETA-CL, are constructed by re-organizing public human attribute datasets. Extensive experiments have been performed on these benchmarks to compare the proposed method with multiple baselines. The superior performance in terms of both recognition accuracies and forgetting ratios demonstrate the effectiveness of the proposed DUAPL for IncPAR.

Progressive Interpretation Synthesis: Interpreting Task Solving by Quantifying Previously Used and Unused Information.

A deep neural network is a good task solver, but it is difficult to make sense of its operation. People have different ideas about how to interpret its operation. We look at this problem from a new perspective where the interpretation of task solving is synthesized by quantifying how much and what previously unused information is exploited in addition to the information used to solve previous tasks. First, after learning several tasks, the network acquires several information partitions related to each task. We propose that the network then learns the minimal information partition that supplements previously learned information partitions to more accurately represent the input. This extra partition is associated with unconceptualized information that has not been used in previous tasks. We manage to identify what unconceptualized information is used and quantify the amount. To interpret how the network solves a new task, we quantify as meta-information how much information from each partition is extracted. We implement this framework with the variational information bottleneck technique. We test the framework with the MNIST and the CLEVR data set. The framework is shown to be able to compose information partitions and synthesize experience-dependent interpretation in the form of meta-information. This system progressively improves the resolution of interpretation upon new experience by converting a part of the unconceptualized information partition to a task-related partition. It can also provide a visual interpretation by imaging what is the part of previously unconceptualized information that is needed to solve a new task.

GopGAN: Gradients Orthogonal Projection Generative Adversarial Network With Continual Learning.

The generative adversarial networks (GANs) in continual learning suffer from catastrophic forgetting. In continual learning, GANs tend to forget about previous generation tasks and only remember the tasks they just learned. In this article, we present a novel conditional GAN, called the gradients orthogonal projection GAN (GopGAN), which updates the weights in the orthogonal subspace of the space spanned by the representations of training examples, and we also mathematically demonstrate its ability to retain the old knowledge about learned tasks in learning a new task. Furthermore, the orthogonal projection matrix for modulating gradients is mathematically derived and its iterative calculation algorithm for continual learning is given so that training examples for learned tasks do not need to be stored when learning a new task. In addition, a task-dependent latent vector construction is presented and the constructed conditional latent vectors are used as the inputs of generator in GopGAN to avoid the disappearance of orthogonal subspace of learned tasks. Extensive experiments on MNIST, EMNIST, SVHN, CIFAR10, and ImageNet-200 generation tasks show that the proposed GopGAN can effectively cope with the issue of catastrophic forgetting and stably retain learned knowledge.

Multi-Object Navigation using Potential Target Position Policy Function.

Visual object navigation is an essential task of embodied AI, which follows the user's demands to let the agent navigate to the goal objects. Previous methods often focus on single object navigation. However, in real life, human demands are generally continuous and multiple, requiring the agent to implement multiple tasks in sequence. These demands can be addressed by repeatedly performing previous single task methods. However, dividing multiple tasks into several independent tasks to perform, without the global optimization between different tasks, the agents' trajectories may overlap, reducing the efficiency of navigation. In this paper, we propose an efficient reinforcement learning framework with a hybrid policy for multi-object navigation, aiming to maximally eliminate the noneffective actions. First, the visual observations are embedded to detect the semantic entities (such as objects). And the detected objects are memorized and projected into semantic maps, which can also be regarded as a long-term memory of the observed environment. Then a hybrid policy consisting of exploration and long-term planning strategies is proposed to predict the potential target position. In particular, when the target is directly oriented, the policy function makes long-term planning to the target based on the semantic map, which is implemented by a sequence of motion actions. In the alternative, when the target is not oriented, the policy function estimates an object potential position towards exploring the most possible objects (positions) that have close relations to the target. The relation between different objects is obtained with prior knowledge, which is used to predict the potential target position by integrating with the memorized semantic map. And then a path to the potential target is planned by the policy function. We evaluate our proposed method on two large-scale 3D realistic environment datasets, Gibson and Matterport3D, and the experimental results demonstrate the effectiveness and generalization of the proposed method.

Reflections on Race and the American Society of International Law

The American Society of International Law (ASIL) is a globally important American professional non-governmental organization, organized by and for international lawyers as a learned society, and influential in its legal interpretations. For its first sixty years, it excluded African Americans. Subsequently, African Americans were allowed incremental but slowly widening participation, though they were still excluded from the American Journal of International Law (AJIL) Board of Editors. Divesting its stock portfolio to oppose apartheid was a reluctant step toward racial justice in 1986. The rise of a new minority-led racial critique of the Society produced, inter alia, the first African American AJIL Board member in 2014, followed by others. It also produced the adoption of the “Richardson Report,” which confirmed the early deliberate African American exclusion, and called for a reparations obligation to Black members. In 2020, new Society statements and priorities toward racial equity and diversity were added to a previous active Black task force (BASIL). However, the Society's transformation toward racial equity will depend upon its committed defiance of the current historic global white nationalist extremism which threatens to penetrate international law in America and the global community. This essay discusses ASIL's racial narratives from its origins of deep racial exclusion, through racial incrementalism to policy changes and reparations obligations toward greater equity, as it now confronts global white identity extremism.

Improved Reinforcement Learning Task Supervisor for Path Planning of Logistics Autonomous System

This study investigates the multi-objective path planning problem in logistics autonomous systems (LAS), where unmanned ground vehicles (UGVs) need to deliver multiple packages to various destinations while avoiding obstacles. Using the null-space-based behavioral control (NSBC) framework, we extend our previous reinforcement learning task supervisor (RLTS) to propose an enhanced RLTS (IRLTS) for optimal path planning and dynamic, simultaneous task priority adjustment. Notably, IRLTS can re-order delivery to minimize total path length when presented with unknown obstacles. Simulations affirm that IRLTS achieves shorter total path lengths than RLTS and outperforms offline optimization-based algorithms on path length and re-planning capability.

Incremental Learning Based on Data Translation and Knowledge Distillation

Recently, deep convolutional neural networks (DCNNs) have achieved remarkable results in image classification tasks. Despite convolutional networks’ great successes, their training process relies on a large amount of data prepared in advance, which is often challenging in real-world applications, such as streaming data and concept drift. For this reason, incremental learning (continual learning) has attracted increasing attention from scholars. However, incremental learning is associated with the challenge of catastrophic forgetting: the performance on previous tasks drastically degrades after learning a new task. In this paper, we propose a new strategy to alleviate catastrophic forgetting when neural networks are trained in continual domains. Specifically, two components are applied: data translation based on transfer learning and knowledge distillation. The former translates a portion of new data to reconstruct the partial data distribution of the old domain. The latter uses an old model as a teacher to guide a new model. The experimental results on three datasets have shown that our work can effectively alleviate catastrophic forgetting by a combination of the two methods aforementioned.