Class-specific Knowledge Research Articles

Decoupled Two-Phase Framework for Class-Incremental Few-Shot Named Entity Recognition

Class-Incremental Few-Shot Named Entity Recognition (CIFNER) aims to identify entity categories that have appeared with only a few newly added (novel) class examples. However, existing class-incremental methods typically introduce new parameters to adapt to new classes and treat all information equally, resulting in poor generalization. Meanwhile, few-shot methods necessitate samples for all observed classes, making them difficult to transfer into a class-incremental setting. Thus, a decoupled two-phase framework method for the CIFNER task is proposed to address the above issues. The whole task is converted to two separate tasks named Entity Span Detection (ESD) and Entity Class Discrimination (ECD) that leverage parameter-cloning and label-fusion to learn different levels of knowledge separately, such as class-generic knowledge and class-specific knowledge. Moreover, different variants, such as the Conditional Random Field-based (CRF-based), word-pair-based methods in ESD module, and add-based, Natural Language Inference-based (NLI-based) and prompt-based methods in ECD module, are investigated to demonstrate the generalizability of the decoupled framework. Extensive experiments on the three Named Entity Recognition (NER) datasets reveal that our method achieves the state-of-the-art performance in the CIFNER setting.

Cross-Domain Few-Shot classification via class-shared and class-specific dictionaries

In Cross-Domain Few-Shot Classification, researchers mainly utilize models which trained with source domain tasks to adapt to the target domain with very few samples, thus causing serious class-difference-caused domain differences. Although researchers have proposed methods to minimize the domain differences, the existing methods have the following drawbacks: 1) most models do not utilize the common knowledge between the source and target domains, and 2) require additional labeled samples from the target domain for finetuning or domain alignment, which is hard to obtain in reality. To address the problem mentioned above, we propose a class-shared and class-specific dictionaries (CSCSD) learning method. To make better utilization of the common knowledge, we apply a class-shared dictionary which is learned to represent the generality of source and target domain. Moreover, class-specific dictionaries are applied to represent the class-specific knowledge that can’t be represented in the class-shared dictionary. Furthermore, unlike most other models, our CSCSD does not require additional target domain samples to meta-train or finetune. With the dictionaries, CSCSD can obtain more distinguishable collaborative representations of samples with the origin representations extracted with the model. To evaluate the effectiveness of CSCSD, we utilize larger datasets, e.g., MiniImageNet and TieredImageNet as source domains and fine-grained datasets, e.g., CUB, Cars, Places, and Plantae as target domains. With our CSCSD, the Cross-Domain Few-Shot accuracy exceeds most domain adaptive Few-Shot which utilizes additional training set in target domains.

Semi-Supervised Unpaired Medical Image Segmentation Through Task-Affinity Consistency.

Deep learning-based semi-supervised learning (SSL) algorithms are promising in reducing the cost of manual annotation of clinicians by using unlabelled data, when developing medical image segmentation tools. However, to date, most existing semi-supervised learning (SSL) algorithms treat the labelled images and unlabelled images separately and ignore the explicit connection between them; this disregards essential shared information and thus hinders further performance improvements. To mine the shared information between the labelled and unlabelled images, we introduce a class-specific representation extraction approach, in which a task-affinity module is specifically designed for representation extraction. We further cast the representation into two different views of feature maps; one is focusing on low-level context, while the other concentrates on structural information. The two views of feature maps are incorporated into the task-affinity module, which then extracts the class-specific representations to aid the knowledge transfer from the labelled images to the unlabelled images. In particular, a task-affinity consistency loss between the labelled images and unlabelled images based on the multi-scale class-specific representations is formulated, leading to a significant performance improvement. Experimental results on three datasets show that our method consistently outperforms existing state-of-the-art methods. Our findings highlight the potential of consistency between class-specific knowledge for semi-supervised medical image segmentation. The code and models are to be made publicly available at https://github.com/jingkunchen/TAC.

Translanguaging as decoloniality-informed knowledge co-construction: a nexus analysis of an English-Medium-Instruction program in China

Abstract Recent studies have shown that translanguaging as a knowledge-construction process can be utilized to decolonize curricula that privilege English-mediated, class-specific knowledge and foster epistemic justice in educational settings. This study focuses on the process whereby translanguaging was mobilized by teachers, tutors, and international students to decolonialize the English-Medium Instruction (EMI) curriculum on an indigenous discipline in a top-rated Chinese university. Drawing on nexus analysis, this study delineates the ways the implemented EMI curriculum was shaped by two processes/practices: (1) intercultural translation as interdiscursive translanguaging, and (2) Chinese philosophy as an embodied way of life, as witnessed in both the teacher-directed interaction order in the classroom as well as in the dialogic interaction order in tutorial sessions, and mediated through the historical bodies of the participants. The analysis reveals that translanguaging provides valuable transknowledging opportunities that encourage negotiations among a plurality of discipline-specific knowledge systems and the construction of decoloniality-informed knowledge production and learning ecologies situated within the unequal geopolitics surrounding knowledge production; this aligns with translanguaging scholars’ decolonial agenda and contributes to developing an operational definition of “transknowledging” based on extant decolonial studies that emphasized the intertwined relationship between transknowledging and translanguaging in decolonializing the EMI curriculum.

Development of a Self-Regulating Evolving Spiking Neural Network for classification problem

This paper presents a new spiking neural network for pattern classification problems, referred to as the Self-Regulating Evolving Spiking Neural (SRESN) classifier, that regulates the learning process of the network. It uses a two layered spiking neural network and the input layer consists of receptive field neurons, which convert a real valued input to spikes using the population coding scheme without any delays. The output layer consists of leaky integrate-and-fire neurons. Since SRESN does not use any delays, the number of network parameters for SRESN is significantly lower than that used by other spiking neural networks, used in this study. During training, the learning algorithm for SRESN, automatically evolves neurons in the output layer based on the training data stream and the current knowledge stored in the network. Depending on the knowledge in the sample and the class specific knowledge stored in the network, it can choose to either add a neuron or update the network parameters or skip learning the sample resulting in self-regulation of the learning process. In case of neuron addition, the weights for the newly added neuron are initialized using a modified rank order scheme which facilitates SRESN for use in online/sequential as well as batch learning modes. The parameter update strategy in SRESN ensures that connections with non-zero postsynaptic potential at the time of the spike are alone updated which helps prevent over training. While evaluating the performance of SRESN, first a study is conducted to assess the impact of various parameters on its performance and establish guidelines to choose suitable values for these parameters. Next, the performance of SRESN, operating in batch mode, is compared with other spiking neural classifiers, including SpikeProp and MuSpiNN, for the UCI benchmark problems of Iris flower classification and Wisconsin breast cancer. Subsequently, the performance of SRESN in online and batch learning mode is compared with an evolving spiking neural classifier for five benchmark data sets from the UCI machine learning repository. Finally, SRESN is applied to solve the practical problem of Epilepsy detection. The performance comparison clearly indicates that SRESN provides a higher generalization accuracy using fewer network parameters.

Hallucination of facial details from degraded images using 3D face models

The goals of this paper are: (1) to enhance the quality of images of faces, (2) to enable 3D Morphable Models (3DMMs) to cope with severely degraded images, and (3) to reconstruct textured 3D faces with details that are not in the input images. Details that are lost in the input images due to blur, low resolution or occlusions, are filled in by the 3DMM and an additional texture enhancement algorithm that adds high-resolution details from example faces. By leveraging class-specific knowledge, this restoration process goes beyond what general image operations such as deblurring or inpainting can achieve. The benefit of the 3DMM for image restoration is that it can be applied to any pose and illumination, unlike image-based methods. However, it is only with the new fitting algorithm that 3DMMs can produce realistic faces from severely degraded images. The new method includes the blurring or downsampling operator explicitly into the analysis-by-synthesis algorithm.

A Metacognitive Complex-Valued Interval Type-2 Fuzzy Inference System

This paper presents a complex-valued interval type-2 neuro-fuzzy inference system (CIT2FIS) and derive its metacognitive projection-based learning (PBL) algorithm. Metacognitive CIT2FIS (Mc-CIT2FIS) consists of a CIT2FIS, which realizes Takagi-Sugeno-Kang type inference mechanism, as its cognitive component. A PBL with self-regulation is its metacognitive component. The rules of CIT2FIS employ interval type-\(2~q\) -Gaussian membership functions that can represent different radial basis functions for different values of \(q\) . As each sample is presented to the network, the metacognitive component monitors the hinge-loss error and class-specific knowledge potential of the current sample to efficiently decide on what-to-learn, when-to-learn, and how-to-learn it. When a new rule is added or existing rules are updated, the optimal parameters of CIT2FIS corresponding to the minimum of the hinge-loss error function are computed using a PBL algorithm derived using the Wirtinger calculus. The performance of Mc-CIT2FIS is evaluated on a set of benchmark real-valued classification problems from the UCI machine learning repository. A circular transformation is used to convert the real-valued features to the complex-valued features in these problems. The performance comparison and statistical study clearly show the superior classification ability of Mc-CIT2FIS. Finally, the proposed complex-valued network is used to solve a practical human action recognition problem that is represented by complex-valued optical flow-based feature set, and a human emotion recognition problem represented using complex-valued Gabor filter-based features. The performance results on these problems substantiate the superior classification ability of Mc-CIT2FIS.

A Metacognitive Neuro-Fuzzy Inference System (McFIS) for Sequential Classification Problems

In this paper, we present a metacognitive sequential learning algorithm for a neuro-fuzzy inference system for classification tasks, which is referred to as a “metacognitive neuro-fuzzy inference system (McFIS).” The McFIS learning algorithm is developed based on the principles of the best human learning strategy, viz., a self-regulatory learning strategy in a metacognitive framework. McFIS has two components: a cognitive component and a metacognitive component. A neuro-fuzzy inference system forms the cognitive component of the McFIS, and a self-regulatory learning mechanism forms its metacognitive component. The learning ability of the cognitive component is monitored and controlled by the self-regulatory learning mechanism. For each sample in the training dataset, the metacognitive component uses its self-adaptive thresholds to choose one of the following learning strategies based on the criteria that depends on class-specific knowledge: 1) sample deletion; 2) sample learning; and 3) sample reserve. Thus, the metacognitive component decides what-to-learn, when-to-learn, and how-to-learn the training samples. When a new rule is added, the parameters of the new rule are assigned such that the rule has minimum overlapping with the adjacent rules as well as the localization property of the Gaussian rules is efficiently exploited. Performance of the McFIS is evaluated using several well-known benchmark multicategory/binary classification datasets from the University of California, Irvine machine learning repository and on a practical human action recognition problem. The results clearly indicate that the proposed metacognitive learning helps the McFIS achieve better performance than other existing classifiers.

Which facial profile do humans expect after seeing a frontal view? a comparison with a linear face model

Manipulated versions of three-dimensional faces that have different profiles, but almost the same appearance in frontal views, provide a novel way to investigate if and how humans use class-specific knowledge to infer depth from images of faces. After seeing a frontal view, participants have to select the profile that matches that view. The profiles are original (ground truth), average, random other, and two solutions computed with a linear face model (3D Morphable Model). One solution is based on 2D vertex positions, the other on pixel colors in the frontal view. The human responses demonstrate that humans neither guess nor just choose the average profile. The results also indicate that humans actually use the information from the front view, and not just rely on the plausibility of the profiles per se. All our findings are perfectly consistent with a correlation-based inference in a linear face model. The results also verify that the 3D reconstructions from our computational algorithms (stimuli 4 and 5) are similar to what humans expect, because they are chosen to be the true profile equally often as the ground-truth profiles. Our experiments shed new light on the mechanisms of human face perception and present a new quality measure for 3D reconstruction algorithms.