Real-world Tasks Research Articles

A Framework for Interpretability in Machine Learning for Medical Imaging.

Interpretability for machine learning models in medical imaging (MLMI) is an important direction of research. However, there is a general sense of murkiness in what interpretability means. Why does the need for interpretability in MLMI arise? What goals does one actually seek to address when interpretability is needed? To answer these questions, we identify a need to formalize the goals and elements of interpretability in MLMI. By reasoning about real-world tasks and goals common in both medical image analysis and its intersection with machine learning, we identify five core elements of interpretability: localization, visual recognizability, physical attribution, model transparency, and actionability. From this, we arrive at a framework for interpretability in MLMI, which serves as a step-by-step guide to approaching interpretability in this context. Overall, this paper formalizes interpretability needs in the context of medical imaging, and our applied perspective clarifies concrete MLMI-specific goals and considerations in order to guide method design and improve real-world usage. Our goal is to provide practical and didactic information for model designers and practitioners, inspire developers of models in the medical imaging field to reason more deeply about what interpretability is achieving, and suggest future directions of interpretability research.

Model-Based Chance-Constrained Reinforcement Learning via Separated Proportional-Integral Lagrangian.

Safety is essential for reinforcement learning (RL) applied in the real world. Adding chance constraints (or probabilistic constraints) is a suitable way to enhance RL safety under uncertainty. Existing chance-constrained RL methods, such as the penalty methods and the Lagrangian methods, either exhibit periodic oscillations or learn an overconservative or unsafe policy. In this article, we address these shortcomings by proposing a separated proportional-integral Lagrangian (SPIL) algorithm. We first review the constrained policy optimization process from a feedback control perspective, which regards the penalty weight as the control input and the safe probability as the control output. Based on this, the penalty method is formulated as a proportional controller, and the Lagrangian method is formulated as an integral controller. We then unify them and present a proportional-integral Lagrangian method to get both their merits with an integral separation technique to limit the integral value to a reasonable range. To accelerate training, the gradient of safe probability is computed in a model-based manner. The convergence of the overall algorithm is analyzed. We demonstrate that our method can reduce the oscillations and conservatism of RL policy in a car-following simulation. To prove its practicality, we also apply our method to a real-world mobile robot navigation task, where our robot successfully avoids a moving obstacle with highly uncertain or even aggressive behaviors.

Transparency challenges in policy evaluation with causal machine learning: improving usability and accountability

Abstract Causal machine learning tools are beginning to see use in real-world policy evaluation tasks to flexibly estimate treatment effects. One issue with these methods is that the machine learning models used are generally black boxes, that is, there is no globally interpretable way to understand how a model makes estimates. This is a clear problem for governments who want to evaluate policy as it is difficult to understand whether such models are functioning in ways that are fair, based on the correct interpretation of evidence and transparent enough to allow for accountability if things go wrong. However, there has been little discussion of transparency problems in the causal machine learning literature and how these might be overcome. This article explores why transparency issues are a problem for causal machine learning in public policy evaluation applications and considers ways these problems might be addressed through explainable AI tools and by simplifying models in line with interpretable AI principles. It then applies these ideas to a case study using a causal forest model to estimate conditional average treatment effects for returns on education study. It shows that existing tools for understanding black-box predictive models are not as well suited to causal machine learning and that simplifying the model to make it interpretable leads to an unacceptable increase in error (in this application). It concludes that new tools are needed to properly understand causal machine learning models and the algorithms that fit them.

COFT-AD: COntrastive Fine-Tuning for Few-Shot Anomaly Detection.

Existing approaches towards anomaly detection (AD) often rely on a substantial amount of anomaly-free data to train representation and density models. However, large anomaly-free datasets may not always be available before the inference stage; in which case an anomaly detection model must be trained with only a handful of normal samples, a.k.a. few-shot anomaly detection (FSAD). In this paper, we propose a novel methodology to address the challenge of FSAD which incorporates two important techniques. Firstly, we employ a model pre-trained on a large source dataset to initialize model weights. Secondly, to ameliorate the covariate shift between source and target domains, we adopt contrastive training to fine-tune on the few-shot target domain data. To learn suitable representations for the downstream AD task, we additionally incorporate cross-instance positive pairs to encourage a tight cluster of the normal samples, and negative pairs for better separation between normal and synthesized negative samples. We evaluate few-shot anomaly detection on 3 controlled AD tasks and 4 real-world AD tasks to demonstrate the effectiveness of the proposed method.

A Clustering-Based Computational Model to Group Students With Similar Programming Skills From Automatic Source Code Analysis Using Novel Features

Throughout a programming course, students develop various source code tasks. Using these tasks to track students' progress can provide clues to the strengths and weaknesses found in each learning topic. This practice allows the teacher to intervene in learning in the first few weeks of class and maximize student gains. However, the biggest challenge is to overcome the amount of work required of the teacher in the manual analysis of all tasks. In this context, our main research objective is to automatically group students with similar programming skills based on the analysis of their submitted source codes. Our research is applied and uses an experimental procedure. First, we prepared the database, with more than 700 real-world source code tasks written in C Language, and distributed it in five different learning topics. Afterward, we define a set of features to be extracted from each learning topic. We defined and extracted 23 features from the source code for five learning topics. Then, we preprocess our database and extract the proposed features. Finally, we grouped the students. After performing the grouping, we obtained four groups of students, which were analyzed using a cluster midpoint calculation. Our results support the monitoring of students throughout the term, offering the teacher the freedom to create new exercises and waiving the obligation of any specific programming environment. We believe that these results can support the teacher in pedagogical decisions closer to the needs of each group of students.

Empowering Real-World Image Super-Resolution With Flexible Interactive Modulation.

Interactive image restoration aims to construct an interactive pathway between users and restoration networks, which empowers users to modulate the restoration results according to their own demands. However, existing methods are primarily limited to training their networks with predefined and simplistic synthetic degradations. Consequently, these methods often encounter significant performance degradation when confronted with real-world degradations that deviate from their assumptions. Furthermore, existing interactive image restoration approaches solely support global modulation, wherein a single modulation factor governs the reconstruction process for the entire image. In this paper, we propose a novel method to perform real-world and intricate image super-resolution in an interactive manner. Specifically, we propose a metric-learning-based degradation estimation strategy to estimate not only the overall degradation level of the entire image but also the finer-grained, pixel- wise degradation within real-world scenarios. This enables local control over the restoration results by selectively modulating the corresponding regions based on the densely-estimated degradation map. Additionally, a new metric-argumented loss is proposed to further enhance the performance of real-world image super-resolution. Through extensive experimentation, we demonstrate the efficacy of our method in achieving exceptional modulation and restoration performance in real-world image super-resolution tasks, all while maintaining an appealing model complexity.

Augmenting Freshman Rural Engineering Students’ Interaction in English Language using Task Based Language Teaching (TBLT)

Abstract— Today’s engineer constantly delivers on both national and international work fronts which gives interaction in English language a key role. This requires teachers to work on their interaction abilities especially for the students who come from rural background. It has been observed that rural students find it difficult to get acclimatized and comfortable with the usage of grammatical concepts in real time situations. This observation necessitates researchers to cater to the needs of these students through proper pedagogical interventions for continuous improvement. Over the years many researchers have used many theories and approaches to enhance interaction skills of rural students but these studies have given temporary solution. Here lies major lacuna thus the current study focuses on how rural freshman engineering students’ interaction can be augmented with the Task based language teaching approach. TBLT is an approach that emphasizes the learning of language through the completion of meaningful tasks. It is based on the belief that language is best learned when it is used to complete a real-world task. It provides students with immersive and engaging learning environment and promotes the practical and effective use of language in real-life situations. The study uses quantitative research approach. A sample of 35 rural freshman students of Technological University is obtained and analyzed. Based on the Pre-Intervention Test and the Post-Intervention Test the results reveal that TBLT strongly augments rural students’ interaction in English language using CEFR’s Common Reference Levels: Self- Assessment Grid. After the intervention, students were requested to evaluate their performance using the assessment grid once again. Encouragingly, there was a clear indication of improvement in their self-assessment outcomes. Keywords—Interaction; Engineers; Pedagogical Intervention; Task Based Language Teaching.

Neural Causal Information Extractor for Unobserved Causes.

Causal inference aims to faithfully depict the causal relationships between given variables. However, in many practical systems, variables are often partially observed, and some unobserved variables could carry significant information and induce causal effects on a target. Identifying these unobserved causes remains a challenge, and existing works have not considered extracting the unobserved causes while retaining the causes that have already been observed and included. In this work, we aim to construct the implicit variables with a generator-discriminator framework named the Neural Causal Information Extractor (NCIE), which can complement the information of unobserved causes and thus provide a complete set of causes with both observed causes and the representations of unobserved causes. By maximizing the mutual information between the targets and the union of observed causes and implicit variables, the implicit variables we generate could complement the information that the unobserved causes should have provided. The synthetic experiments show that the implicit variables preserve the information and dynamics of the unobserved causes. In addition, extensive real-world time series prediction tasks show improved precision after introducing implicit variables, thus indicating their causality to the targets.

Language education, digital citizenship, and Juri Lotman’s semiosphere

The present paper explores the use of language teaching for educating students into becoming digitally-literate citizens, as well as conscious, active members of the digital universe. The material discussed derives from the eLang project, a flagship project of the European Centre for Modern Languages. The guidelines, and real-world and reflexive tasks put together by the eLang team of experts, along with the theoretical framework employed are examined with respect to the notion of semiosphere by Juri M. Lotman, and the way this endorses digital transformation in language education. Seen, thus, as partaking in the digital semiosphere, and at the same time in the multiplicity of the semiotic systems ingrained in it, the current language student and future citizen assumes different roles, interacts with distinct as well as overlapping communities, is asked to make sense of multimodal resources, so as to eventually acquire far more than the skills of a digital user. The eLang material addresses, in this way, the demand for training language students in the multifarious literacies that digital literacy has come to encompass. The language student overrides, thus, classroom topography and the boundaries of conventional language education, and traces those of the globalised digital semiosphere, within which it cannot but soar.

Amplifying pathological detection in EEG signaling pathways through cross-dataset transfer learning

Pathology diagnosis based on EEG signals and decoding brain activity holds immense importance in understanding neurological disorders. With the advancement of artificial intelligence methods and machine learning techniques, the potential for accurate data-driven diagnoses and effective treatments has grown significantly. However, applying machine learning algorithms to real-world datasets presents diverse challenges at multiple levels. The scarcity of labeled data, especially in low regime scenarios with limited availability of real patient cohorts due to high costs of recruitment, underscores the vital deployment of scaling and transfer learning techniques. In this study, we explore a real-world pathology classification task to highlight the effectiveness of data and model scaling and cross-dataset knowledge transfer. As such, we observe varying performance improvements through data scaling, indicating the need for careful evaluation and labeling. Additionally, we identify the challenges of possible negative transfer and emphasize the significance of some key components to overcome distribution shifts and potential spurious correlations and achieve positive transfer. We see improvement in the performance of the target model on the target (NMT) datasets by using the knowledge from the source dataset (TUAB) when a low amount of labeled data was available. Our findings demonstrated that a small and generic model (e.g. ShallowNet) performs well on a single dataset, however, a larger model (e.g. TCN) performs better in transfer learning when leveraging a larger and more diverse dataset.

Dense depth distillation with out-of-distribution simulated images

We study data-free knowledge distillation (KD) for monocular depth estimation (MDE), which learns a lightweight model for real-world depth perception tasks by compressing it from a trained teacher model while lacking training data in the target domain. Owing to the essential difference between image classification and dense regression, previous methods of data-free KD are not applicable to MDE. To strengthen its applicability in real-world tasks, in this paper, we propose to apply KD with out-of-distribution simulated images. The major challenges to be resolved are (i) lacking prior information about scene configurations of real-world training data and (ii) domain shift between simulated and real-world images. To cope with these difficulties, we propose a tailored framework for depth distillation. The framework generates new training samples for embracing a multitude of possible object arrangements in the target domain and utilizes a transformation network to efficiently adapt them to the feature statistics preserved in the teacher model. Through extensive experiments on various depth estimation models and two different datasets, we show that our method outperforms the baseline KD by a good margin and even achieves slightly better performance with as few as 1/6 of training images, demonstrating a clear superiority.

Anodal cerebellar t-DCS impacts skill learning and transfer on a robotic surgery training task

The cerebellum has demonstrated a critical role during adaptation in motor learning. However, the extent to which it can contribute to the skill acquisition of complex real-world tasks remains unclear. One particularly challenging application in terms of motor activities is robotic surgery, which requires surgeons to complete complex multidimensional visuomotor tasks through a remotely operated robot. Given the need for high skill proficiency and the lack of haptic feedback, there is a pressing need for understanding and improving skill development. We investigated the effect of cerebellar transcranial direct current stimulation applied during the execution of a robotic surgery training task. Study participants received either real or sham stimulation while performing a needle driving task in a virtual (simulated) and a real-world (actual surgical robot) setting. We found that cerebellar stimulation significantly improved performance compared to sham stimulation at fast (more demanding) execution speeds in both virtual and real-world training settings. Furthermore, participants that received cerebellar stimulation more effectively transferred the skills they acquired during virtual training to the real world. Our findings underline the potential of non-invasive brain stimulation to enhance skill learning and transfer in real-world relevant tasks and, more broadly, its potential for improving complex motor learning.

The increase of writing proficiency within task-based language extended writing integrating self-regulated learning and technological facilitation

This research dives into the complex world of improving writing skills using Task-Based Language Teaching approach (TBLT). It looks at the details of writing skills for intermediate-level students, focusing on the combination of self-learning strategies and the use of technology. Using modern teaching ideas, the study points out the difficulties students face when trying to improve their writing skills. It outlines the important features of good writing, which include clearness, logical order, good grammar, and the ability to engage readers. In addition, the study points out specific skills that are important for good writing, such as understanding grammar, organizing ideas, creating content, and understanding the writing process. The research also highlights the important role of TBLT in improving writing skills. TBLT's focus on real-world tasks and communication-based learning environments is seen as key in improving language skills, including writing. The study looks at how TBLT aligns with self-learning principles, showing its potential to help students control their own learning process, increasing their motivation, engagement, and ownership. The study also emphasizes the significant impact of technology on teaching writing within TBLT. The combination of the digital tools within TBLT creates an environment that encourages self-learning and proficient English creative writers.

Interpretable Imitation Learning with Symbolic Rewards

Sample inefficiency of deep reinforcement learning methods is a major obstacle for their use in real-world tasks as they naturally feature sparse rewards. In fact, this from-scratch approach is often impractical in environments where extreme negative outcomes are possible. Recent advances in imitation learning have improved sample efficiency by leveraging expert demonstrations. Most work along this line of research employs neural network-based approaches to recover an expert cost function. However, the complexity and lack of transparency make neural networks difficult to trust and deploy in the real world. In contrast, we present a method for extracting interpretable symbolic reward functions from expert data, which offers several advantages. First, the learned reward function can be parsed by a human to understand, verify and predict the behavior of the agent. Second, the reward function can be improved and modified by an expert. Finally, the structure of the reward function can be leveraged to extract explanations that encode richer domain knowledge than standard scalar rewards. To this end, we use an autoregressive recurrent neural network that generates hierarchical symbolic rewards represented by simple symbolic trees. The recurrent neural network is trained via risk-seeking policy gradients. We test our method in MuJoCo environments as well as a chemical plant simulator. We show that the discovered rewards can significantly accelerate the training process and achieve similar or better performance than neural network-based algorithms.

Neuro-symbolic recommendation model based on logic query

Recommendation systems assist users in finding items that are relevant to them. However, recommendation is not only an inductive statistics problem using data; it is also a cognitive task of reasoning decisions based on knowledge extracted from information. Hence, a logic system can naturally be incorporated for reasoning in a recommendation task. Although logic-based hard-rule approaches can provide a powerful reasoning ability, they struggle to cope with inconsistent and incomplete knowledge in real-world tasks, especially for complex tasks such as recommendation. To address these issues, we propose a neuro-symbolic recommendation model, which transforms user history interactions into a logic expression and then transforms the recommendation prediction into a query task based on this logic expression. The logic expressions are then computed based on the modular logic operations of the neural network. We also construct an implicit logic encoder to reasonably reduce the complexity of the logic computation. Finally, a user’s interest items can be queried in the vector space based on the computation results. Experiments on three well-known datasets verified that the proposed method outperforms state of the art shallow, deep, session, and reasoning models.

Personalized and privacy-preserving federated heterogeneous medical image analysis with PPPML-HMI

Heterogeneous data is endemic due to the use of diverse models and settings of devices by hospitals in the field of medical imaging. However, there are few open-source frameworks for federated heterogeneous medical image analysis with personalization and privacy protection without the demand to modify the existing model structures or to share any private data. Here, we proposed PPPML-HMI, a novel open-source learning paradigm for personalized and privacy-preserving federated heterogeneous medical image analysis. To our best knowledge, personalization and privacy protection were discussed simultaneously for the first time under the federated scenario by integrating the PerFedAvg algorithm and designing the novel cyclic secure aggregation with the homomorphic encryption algorithm. To show the utility of PPPML-HMI, we applied it to a simulated classification task namely the classification of healthy people and patients from the RAD-ChestCT Dataset, and one real-world segmentation task namely the segmentation of lung infections from COVID-19 CT scans. Meanwhile, we applied the improved deep leakage from gradients to simulate adversarial attacks and showed the strong privacy-preserving capability of PPPML-HMI. By applying PPPML-HMI to both tasks with different neural networks, a varied number of users, and sample sizes, we demonstrated the strong generalizability of PPPML-HMI in privacy-preserving federated learning on heterogeneous medical images.

Evolving interpretable decision trees for reinforcement learning

In recent years, reinforcement learning (RL) techniques have achieved great success in many different applications. However, their heavy reliance on complex deep neural networks makes most RL models uninterpretable, limiting their application in domains where trust and security are important. To address this challenge, we propose MENS-DT-RL, an algorithm capable of constructing interpretable models for RL via the evolution of decision tree (DT) models. MENS-DT-RL uses a multi-method ensemble algorithm to evolve univariate DTs, guiding the process with a fitness metric that prioritizes interpretability and consistent high performance. Three different initializations for the MENS-DT-RL are proposed, including the use of Imitation Learning (IL) techniques, and a novel pruning approach that reduces solution size without compromising performance. To evaluate the proposed approach, we compare it with other models from the literature on three benchmark tasks from the OpenAI Gym library, as well as on a fertilization problem inspired by real-world crop management. To the best of our knowledge, the proposed scheme is the first to solve the Lunar Lander benchmark with both interpretability and a high confidence rate (90% of episodes are successful), as well as the first to solve the Mountain Car environment with a tree of only 7 nodes. On the real-world task, the proposed MENS-DT-RL is able to produce solutions with the same quality as deep RL policies, with the added bonus of interpretability. We also analyze the best solutions found by the algorithm and show that they are not only interpretable but also diverse in their behavior, empowering the end user with the choice of which model to apply. Overall, the findings show that the proposed approach is capable of producing high-quality transparent models for RL, achieving interpretability without losing performance.

Adaptive Adversarial Contrastive Learning for Cross-Domain Recommendation

Graph-based cross-domain recommendations (CDRs) are useful for suggesting appropriate items because of their promising ability to extract features from user–item interactions and transfer knowledge across domains. Thus, the model can effectively alleviate cold start and data sparsity issues. Although the graph-based CDRs can capture valuable information, they still have some limitations. First, embeddings are highly vulnerable to noisy interactions, because the message aggregation in the graph convolutional network can further enlarge the impact. Second, because of the property of graph-structured data, the influence of high-degree nodes on representation learning is more than that of the long-tail items, and this can cause a poor recommendation performance. In this study, we devised a novel A daptive A dversarial C ontrastive L earning framework for graph-based C ross- D omain R ecommendation ( ACLCDR ). The ACLCDR introduces reinforcement learning to generate adaptive augmented samples for contrastive learning tasks. Then, we leveraged a multitask training strategy to jointly optimize the model with auxiliary tasks. Finally, we verified the effectiveness of the ACLCDR through nine real-world cross-domain tasks adopted from Amazon and Douban. We observed that ACLCDR exceeded the best state-of-the-art baseline by 25%, 42.5%, 16.3%, and 23.8% in terms of HR@ 10 and NDCG@10 for the Music & Movie task from Amazon.

Mitigating the influence of domain shift in skin lesion classification: A benchmark study of unsupervised domain adaptation methods

The potential of deep neural networks in skin lesion classification has already been demonstrated to be on-par if not superior to the dermatologists’ diagnosis in experimental settings. However, the performance of these models usually deteriorates in real-world scenarios, where the test data differs significantly from the training data (i.e. domain shift). This concerning limitation for models intended to be used in real-world skin lesion classification tasks poses a risk to patients. For example, different image acquisition systems or previously unseen anatomical sites on the patient can suffice to cause such domain shifts. Mitigating the negative effect of such shifts is therefore crucial, but developing effective methods to address domain shift has proven to be challenging. In this study, we carry out a comparative analysis of eight different unsupervised domain adaptation methods to analyze their effectiveness in improving generalization for dermoscopic datasets. To ensure robustness of our findings, we test each method on a total of ten derived datasets, thereby covering a variety of possible domain shifts. In addition, we investigated which factors in the domain shifted datasets have an impact on the effectiveness of domain adaptation methods. Our findings show that all of the eight domain adaptation methods result in improved AUPRC for the majority of analyzed datasets. Altogether, these results indicate that unsupervised domain adaptations generally lead to performance improvements for the binary melanoma-nevus classification task regardless of the nature of the domain shift. However, small or heavily imbalanced datasets lead to a reduced conformity of the results due to the influence of these factors on the methods’ performance.

Spatial-temporal knowledge distillation for lightweight network traffic anomaly detection

Deep learning-based network traffic anomaly detection methods have been the mainstream approaches to enhancing the accuracy performance of Network Intrusion Detection Systems (NIDSs). However, there are several problems that remain to be addressed in practical scenarios. First, the memory and computing power of intelligent terminals restrict the deployment of computationally intensive deep learning methods. Second, the depth and width of representations are of central significance for the accuracy of detection, at the cost of memory consumption and computational complexity. Third, the long tail effect spawned by the category imbalance of network traffic is prevalent in real-world fine-grained anomaly detection tasks. Therefore, we propose a Spatial-Temporal Knowledge Distillation (STKD) algorithm framework for lightweight network traffic anomaly detection to tackle the challenges. Integrating multi-scale One-Dimensional Convolutional Neural Network (1D CNN) and Long Short-Term Memory Network (LSTM), and adopting identity mapping, we propose a Multi-Scale Spatial-Temporal Residual Network (MSSTRNet) as the teacher model for deep spatial-temporal feature extraction of network traffic. Based on Knowledge Distillation (KD), we compress MSSTRNet to the lightweight student model named LENet which is suitable for deployment. Introducing Focal Loss (FL) instead of Cross Entropy (CE) Loss into the KD process, we attempt to alleviate the long tail effect in the fine-grained anomaly detection tasks. Experiments demonstrate the superiority of our proposed methods on accuracy performance, memory consumption and computation complexity.