Specific Problem Domain Research Articles

Detection of Fake Instagram Accounts via Machine Learning Techniques

This paper focuses on the detection of fake accounts on Instagram and proposes a novel solution that aims to address this problem. More specifically, a machine learning-based solution is introduced that can be employed by Instagram-based applications to combat this phenomenon. To accomplish this, publicly available data from Instagram users are collected and processed. After making the necessary feature additions to and removals from these data, they are fed into machine learning algorithms with the aim of detecting fake Instagram accounts. The interest of the results presented in this study lies in their comparability with the results of other similar studies, with the difference that fewer input features have been used. This addresses significant research challenges regarding the performance of and confidence in obtained results in this specific problem domain.

Parametric picture fuzzy cross-entropy measures based on [formula omitted]-Choquet integral for building material recognition

This research paper introduces a novel picture fuzzy cross-entropy measure that utilizes Lin’s divergence as a basis for comparing two picture fuzzy sets. To improve the flexibility and applicability of this new cross-entropy measure, a family of parametric cross-entropy measures is defined. By adjusting the parameters in applications, the influence of the degree of positive membership, negative membership, and neutral membership on the picture fuzzy cross-entropy can be observed, allowing for a more tailored analysis. Additionally, we investigate the relationship between power weighted means and the d-Choquet integral, which serves as an extension of the ordinary Choquet integral. Using this knowledge, picture fuzzy cross-entropy measures based on the d-Choquet integral are introduced to yield more sensitive results, particularly when interactions between criteria exist in specific problem domains. This consideration of criterion interactions is often absent in existing cross-entropy measures. Moreover, an algorithm is presented for solving pattern recognition problems and applied to a building material recognition problem sourced from existing literature. Then, we proposed another algorithm and use it to investigate a novel material classification problem. Through these applications, the effectiveness of the proposed cross-entropy measures in pattern recognition is demonstrated. The paper conducts a comparative analysis between existing methods and the proposed approaches, followed by a sensitivity analysis. This analysis involves manipulating the parameters derived from both the parametric cross-entropy measures and the d-Choquet integrals to assess their respective impacts and sensitivities. Finally, the results regarding the classification problem are examined with performance metrics such as accuracy, precision, recall, and F1 score.

Recommendations for Improving End-User Programming Education: A Case Study with Undergraduate Chemistry Students.

Programming is widespread in multiple domains and is being integrated into various discipline-specific university courses where, like students in a typical introductory computing course, students from other disciplines face challenges with learning to program. We offer a case study in which we study undergraduate students majoring in either chemistry or biochemistry as they learn programming in a physical chemistry course sequence. Using surveys and think-aloud sessions with students, we conducted a thematic content analysis to explain the challenges they face in this endeavor. We found that students struggled to transfer their programming knowledge to new representations and problems, and they did not have strategies in place for solving problems with programming. These facts combine to lower students' confidence in their programming abilities, making it less likely that they will reach for computing to help solve domain-specific problems. We recommend that students in end-user programming contexts be explicitly taught the skills of abstraction, decomposition, and metacognitive awareness as they pertain to programming.

Improving medical reasoning through retrieval and self-reflection with retrieval-augmented large language models.

Recent proprietary large language models (LLMs), such as GPT-4, have achieved a milestone in tackling diverse challenges in the biomedical domain, ranging from multiple-choice questions to long-form generations. To address challenges that still cannot be handled with the encoded knowledge of LLMs, various retrieval-augmented generation (RAG) methods have been developed by searching documents from the knowledge corpus and appending them unconditionally or selectively to the input of LLMs for generation. However, when applying existing methods to different domain-specific problems, poor generalization becomes apparent, leading to fetching incorrect documents or making inaccurate judgments. In this paper, we introduce Self-BioRAG, a framework reliable for biomedical text that specializes in generating explanations, retrieving domain-specific documents, and self-reflecting generated responses. We utilize 84k filtered biomedical instruction sets to train Self-BioRAG that can assess its generated explanations with customized reflective tokens. Our work proves that domain-specific components, such as a retriever, domain-related document corpus, and instruction sets are necessary for adhering to domain-related instructions. Using three major medical question-answering benchmark datasets, experimental results of Self-BioRAG demonstrate significant performance gains by achieving a 7.2% absolute improvement on average over the state-of-the-art open-foundation model with a parameter size of 7B or less. Similarly, Self-BioRAG outperforms RAG by 8% Rouge-1 score in generating more proficient answers on two long-form question-answering benchmarks on average. Overall, we analyze that Self-BioRAG finds the clues in the question, retrieves relevant documents if needed, and understands how to answer with information from retrieved documents and encoded knowledge as a medical expert does. We release our data and code for training our framework components and model weights (7B and 13B) to enhance capabilities in biomedical and clinical domains. Self-BioRAG is available at https://github.com/dmis-lab/self-biorag.

A Clustering and PL/SQL-Based Method for Assessing MLP-Kmeans Modeling

With new high-performance server technology in data centers and bunkers, optimizing search engines to process time and resource consumption efficiently is necessary. The database query system, upheld by the standard SQL language, has maintained the same functional design since the advent of PL/SQL. This situation is caused by recent research focused on computer resource management, encryption, and security rather than improving data mining based on AI tools, machine learning (ML), and artificial neural networks (ANNs). This work presents a projected methodology integrating a multilayer perceptron (MLP) with Kmeans. This methodology is compared with traditional PL/SQL tools and aims to improve the database response time while outlining future advantages for ML and Kmeans in data processing. We propose a new corollary: hk→H=SSE(C),wherek>0and∃X, executed on application software querying data collections with more than 306 thousand records. This study produced a comparative table between PL/SQL and MLP-Kmeans based on three hypotheses: line query, group query, and total query. The results show that line query increased to 9 ms, group query increased from 88 to 2460 ms, and total query from 13 to 279 ms. Testing one methodology against the other not only shows the incremental fatigue and time consumption that training brings to database query but also that the complexity of the use of a neural network is capable of producing more precision results than the simple use of PL/SQL instructions, and this will be more important in the future for domain-specific problems.

A domain-independent agent architecture for adaptive operation in evolving open worlds

1Angry Birds Competition. Model-based reasoning agents are ill-equipped to act in novel situations in which their model of the environment no longer sufficiently represents the world. We propose HYDRA, a framework for designing model-based agents operating in mixed discrete-continuous worlds that can autonomously detect when the environment has evolved from its canonical setup, understand how it has evolved, and adapt the agents' models to perform effectively. HYDRA is based upon PDDL+, a rich modeling language for planning in mixed, discrete-continuous environments. It augments the planning module with visual reasoning, task selection, and action execution modules for closed-loop interaction with complex environments. HYDRA implements a novel meta-reasoning process that enables the agent to monitor its own behavior from a variety of aspects. The process employs a diverse set of computational methods to maintain expectations about the agent's own behavior in an environment. Divergences from those expectations are useful in detecting when the environment has evolved and identifying opportunities to adapt the underlying models. HYDRA builds upon ideas from diagnosis and repair and uses a heuristics-guided search over model changes such that they become competent in novel conditions. The HYDRA framework has been used to implement novelty-aware agents for three diverse domains - CartPole++ (a higher dimension variant of a classic control problem), Science Birds (an IJCAI competition problem1), and PogoStick (a specific problem domain in Minecraft). We report empirical observations from these domains to demonstrate the efficacy of various components in the novelty meta-reasoning process.

Large language models present new questions for decision support

Large language models (LLMs) have proven capable of assisting with many aspects of organizational decision making, such as helping to collect information from databases and helping to brainstorm possible courses of action ahead of making a choice. We propose that broad adoption of these technologies introduces new questions in the study of decision support systems, which assist people with complex and open-ended choices in business. Where traditional study of decision support has focused on bespoke tools to solve narrow problems in specific domains, LLMs offer a general-purpose decision support technology which can be applied in many contexts. To organize the wealth of new questions which result from this shift, we turn to a classic framework from Herbert Simon, which proposes that decision making requires collecting evidence, considering alternatives, and finally making a choice. Working from Simon’s framework, we describe how LLMs introduce new questions at each stage of this decision-making process. We then group new questions into three overarching themes for future research, centered on how LLMs will change individual decision making, how LLMs will change organizational decision making, and how to design new decision support technologies which make use of the new capabilities of LLMs.

Activation Functions for Neural Networks: Application and Performance-based Comparison

Past decade has seen explosive growth of Deep Learning (DL) algorithms based on Artificial Neural Networks (ANNs) and its applications in vast emerging domains to solve real world complex problems. The DL architecture uses Activation Functions (AFs), to perform the task of finding relationship between the input feature and the output. Essential building blocks of any ANN are AFs which bring the required non-linearity of the output in the Output layer of network. Layers of ANNs are combinations of linear and nonlinear AFs. Most extensively used AFs are Sigmoid, Hyperbolic Tangent (Tanh), Rectified Linear Unit (ReLU) etc to name a few. Choosing an AF for a particular AF depends on various factors such as Nature of Application, Design of ANN, Optimizers used in the network, Complexity of Data etc. This paper presents a survey on most widely used AFs along with the important consideration while selecting an AF on a specific problem domain. A broad guideline on selecting an AF based on the literature survey has been presented to help researchers in employing suitable AF in their problem domain.

Assessing causal links between age at menarche and adolescent mental health: a Mendelian randomisation study

BackgroundThe timing of puberty may have an important impact on adolescent mental health. In particular, earlier age at menarche has been associated with elevated rates of depression in adolescents. Previous research suggests that this relationship may be causal, but replication and an investigation of whether this effect extends to other mental health domains is warranted.MethodsIn this Registered Report, we triangulated evidence from different causal inference methods using a new wave of data (N = 13,398) from the Norwegian Mother, Father, and Child Cohort Study. We combined multiple regression, one- and two-sample Mendelian randomisation (MR), and negative control analyses (using pre-pubertal symptoms as outcomes) to assess the causal links between age at menarche and different domains of adolescent mental health.ResultsOur results supported the hypothesis that earlier age at menarche is associated with elevated depressive symptoms in early adolescence based on multiple regression (β = − 0.11, 95% CI [− 0.12, − 0.09], pone-tailed < 0.01). One-sample MR analyses suggested that this relationship may be causal (β = − 0.07, 95% CI [− 0.13, 0.00], pone-tailed = 0.03), but the effect was small, corresponding to just a 0.06 standard deviation increase in depressive symptoms with each earlier year of menarche. There was also some evidence of a causal relationship with depression diagnoses during adolescence based on one-sample MR (OR = 0.74, 95% CI [0.54, 1.01], pone-tailed = 0.03), corresponding to a 29% increase in the odds of receiving a depression diagnosis with each earlier year of menarche. Negative control and two-sample MR sensitivity analyses were broadly consistent with this pattern of results. Multivariable MR analyses accounting for the genetic overlap between age at menarche and childhood body size provided some evidence of confounding. Meanwhile, we found little consistent evidence of effects on other domains of mental health after accounting for co-occurring depression and other confounding.ConclusionsWe found evidence that age at menarche affected diagnoses of adolescent depression, but not other domains of mental health. Our findings suggest that earlier age at menarche is linked to problems in specific domains rather than adolescent mental health in general.

General fractal dimensions of graphs of products and sums of continuous functions and their decompositions

This study takes a broad approach to the fractal geometry problem and proposes an intrinsic definition of the general box dimensions and the general Hausdorff and packing dimensions by taking into account sums of the type ∑ih−1(sg(ri)) for some prescribed functions h, g and for all positive real s. Our primary aim is to conduct a more comprehensive exploration of the fractal dimensions exhibited by graphs resulting from the products and sums of two continuous functions. This article is entirely devoted to the investigation of this specific problem domain. In the course of our research, we furnish explicit formulas for computing both the general upper box dimension and the general lower box dimension associated with the graphs formed by the products and sums of two continuous functions. Additionally, we shed light on the decomposition of continuous functions into the sum of two continuous functions, employing the framework of the general lower and upper box dimensions. Furthermore, we establish an upper bound for the general upper box dimension applicable to each element within a finite ring of polynomials derived from continuous functions defined over the real number field R. Our inquiry also extends to the examination of general Hausdorff and packing dimensions for the graphs generated through both the summation and multiplication of two continuous functions.

Applied Data Analysis: A Problem-based Learning Approach

This paper examines the transition of a conventional multivariate statistics module to a problem-based learning module, first implemented in 2021. The primary objective was to enhance students’ problem-solving skills, bridging the gap between mathematical concepts and real-world applications. The approach was implemented to instil a deeper understanding of real-world data analysis, emphasising the interpretation of domain specific problems in mathematical terms and the production of reports for industrial stakeholders.Findings indicate that the integration of problem-based methods not only improved students’ comprehension of statistical techniques but also fostered a more profound appreciation for their practical utility in diverse professional contexts. The problem-solving cycle, a central component of the approach, guided students in critically analysing complex challenges and formulating data driven solutions. Furthermore, this study emphasises the potential for replicating the industrial study group experience within an undergraduate teaching environment.Adopting a problem-based learning approach in the teaching of data analysis empowers students to apply their analytical skills effectively to real-world scenarios, strengthening their capacity to communicate insights and solutions to industrial stakeholders. The study underscores the value of aligning educational practices with the demands of data-driven industries, providing students with a competitive advantage in future research and the job market. The study is descriptive and reflective in nature.

Optimizing ADMM and Over-Relaxed ADMM Parameters for Linear Quadratic Problems

The Alternating Direction Method of Multipliers (ADMM) has gained significant attention across a broad spectrum of machine learning applications. Incorporating the over-relaxation technique shows potential for enhancing the convergence rate of ADMM. However, determining optimal algorithmic parameters, including both the associated penalty and relaxation parameters, often relies on empirical approaches tailored to specific problem domains and contextual scenarios. Incorrect parameter selection can significantly hinder ADMM's convergence rate. To address this challenge, in this paper we first propose a general approach to optimize the value of penalty parameter, followed by a novel closed-form formula to compute the optimal relaxation parameter in the context of linear quadratic problems (LQPs). We then experimentally validate our parameter selection methods through random instantiations and diverse imaging applications, encompassing diffeomorphic image registration, image deblurring, and MRI reconstruction.

Connected Components for Scaling Partial-order Blocking to Billion Entities

In entity resolution, blocking pre-partitions data for further processing by more expensive methods. Two entity mentions are in the same block if they share identical or related blocking-keys . Previous work has sometimes related blocking keys by grouping or alphabetically sorting them, but—as was shown for author disambiguation—the respective equivalences or total orders are not necessarily well-suited to model the logical matching-relation between blocking keys. To address this, we present a novel blocking approach that exploits the subset partial order over entity representations to build a matching-based bipartite graph, using connected components as blocks. To prevent over- and underconnectedness, we allow specification of overly general and generalization of overly specific representations. To build the bipartite graph, we contribute a new parallellized algorithm with configurable time/space tradeoff for minimal element search in the subset partial order. As a job-based approach, it combines dynamic scalability and easier integration to make it more convenient than the previously described approaches. Experiments on large gold standards for publication records, author mentions, and affiliation strings suggest that our approach is competitive in performance and allows better addressing of domain-specific problems. For duplicate detection and author disambiguation, our method offers the expected performance as defined by the vector-similarity baseline used in another work on the same dataset and the common surname, first-initial baseline. For top-level institution resolution, we have reproduced the challenges described in prior work, strengthening the conclusion that for affiliation data, overlapping blocks under minimal elements are more suitable than connected components.

Neural network crossover in genetic algorithms using genetic programming

The use of genetic algorithms (GAs) to evolve neural network (NN) weights has risen in popularity in recent years, particularly when used together with gradient descent as a mutation operator. However, crossover operators are often omitted from such GAs as they are seen as being highly destructive and detrimental to the performance of the GA. Designing crossover operators that can effectively be applied to NNs has been an active area of research with success limited to specific problem domains. The focus of this study is to use genetic programming (GP) to automatically evolve crossover operators that can be applied to NN weights and used in GAs. A novel GP is proposed and used to evolve both reusable and disposable crossover operators to compare their efficiency. Experiments are conducted to compare the performance of GAs using no crossover operator or a commonly used human designed crossover operator to GAs using GP evolved crossover operators. Results from experiments conducted show that using GP to evolve disposable crossover operators leads to highly effectively crossover operators that significantly improve the results obtained from the GA.

Hybrid whale optimized crow search algorithm and multi-SVM classifier for effective system level test case selection

A novel approach to enhance software testing through intelligent test case selection is proposed in this work. The proposed method combines feature extraction, clustering, and a hybrid optimization algorithm to improve testing effectiveness while reducing resource overhead. It employs a context encoder to extract relevant features from software code, enhancing the accuracy of subsequent testing. Through the use of Fuzzy C-means (FCM) clustering, the test cases are classified into groups, streamlining the testing process by identifying similar cases. To optimize feature selection, a Hybrid Whale Optimized Crow Search Algorithm (HWOCSA), which intelligently combines the strengths of both Whale Optimization Algorithm (WOA) and Crow Search Algorithm (CSA) is introduced. This hybrid approach mitigates limitations while maximizing the selection of pertinent features for testing. The ultimate contribution of this work lies in the proposal of a multi-SVM classifier, which refines the test case selection process. Each classifier learns specific problem domains, generating predictions that guide the selection of test cases with unprecedented precision. Experimental results demonstrate that the proposed method achieves remarkable improvements in testing outcomes, including enhanced performance metrics, reduced computation time, and minimized training data requirements. By significantly streamlining the testing process and accurately selecting relevant test cases, this work paves the way for higher quality software updates at a reduced cost.

Research On Pruning Methods for Mobilenet Convolutional Neural Network

This paper comprehensively reviews pruning methods for MobileNet convolutional neural networks. MobileNet is a lightweight convolutional neural network suitable for resource-constrained environments such as mobile devices.Various pruning methods can be applied to reduce the model's storage space and computational complexity, including channel pruning, kernel pruning, and weight pruning. Channel pruning removes unimportant channels to reduce redundant parameters and computations in the model, while kernel pruning reduces redundant calculations by pruning convolutional kernels. Weight pruning involves setting small-weighted elements to zero to remove unimportant weights. These pruning methods can be used individually or in combination. After pruning, fine-tuning is necessary to restore the model's performance. Factors such as pruning rate, pruning order, and pruning location need to be considered to achieve a balance between reducing model size and computational complexity while minimizing performance loss. Pruning methods based on MobileNet convolutional neural networks reduce the parameter count and computational complexity, improving model lightweightness and inference efficiency. These methods are of significant value in resource-constrained environments such as mobile devices. This review provides insights into pruning methods for MobileNet convolutional neural networks and their applications in lightweight and efficient model deployment. Further advancements such as automated pruning methods driven by reinforcement learning algorithms can enhance the pruning process to achieve optimal model compression effects. Future research should focus on adapting and optimizing these pruning methods for specific problem domains and achieving even higher compression ratios and computational speedups.

Equitable Machine Learning for Hypoglycaemia Risk Management.

We developed a machine learning (ML) model for the detection of patients with high risk of hypoglycaemic events during their hospital stay to improve the detection and management of hypoglycaemia. Our model was trained on data from a regional local health care district in Australia. The model was found to have good predictive performance in the general case (AUC 0.837). We conducted subgroup analysis to ensure that the model performed in a way that did not disadvantage population subgroups, in this case based on gender or indigenous status. We found that our specific problem domain assisted us in reducing unwanted bias within the model, because it did not rely on practice patterns or subjective judgements for the outcome measure. With careful analysis for equity there is great potential for ML models to automate the detection of high-risk cohorts and automate mitigation strategies to reduce preventable errors.

Object Tracking with Sensor Fusion – An Interactive Learning Tool

Body tracking plays a key role in autonomous navigation applications. Behavior that resists inertia can be modelled as a dynamical system, wherein the kinematic component is constituted by the action of motion. Such a system may then be subjected to estimation algorithms and control laws formulated by systems theory, according to the specific problem domain for which it is modelled. This paper presents a detailed comparison of three main statistical algorithms for estimating dynamical system parameters: the linear, extended, and unscented Kalman filters. The body motion is intercepted by sensor fusion. To facilitate visual validation and concretization of the theoretical notions presented, a two-dimensional (2D) game-like graphical application has been developed to enhance user comprehension.

Building Domain-Specific Machine Learning Workflows: A Conceptual Framework for the State-of-the-Practice

Domain experts are increasingly employing machine learning to solve their domain-specific problems. This article presents to software engineering researchers the six key challenges that a domain expert faces in addressing their problem with a computational workflow, and the underlying executable implementation. These challenges arise out of our conceptual framework which presents the “route” of transformations that a domain expert may choose to take while developing their solution. To ground our conceptual framework in the state-of-the-practice, this article discusses a selection of available textual and graphical workflow systems and their support for the transformations described in our framework. Example studies from the literature in various domains are also examined to highlight the tools used by the domain experts as well as a classification of the domain-specificity and machine learning usage of their problem, workflow, and implementation. The state-of-the-practice informs our discussion of the six key challenges, where we identify which challenges and transformations are not sufficiently addressed by available tools. We also suggest possible research directions for software engineering researchers to increase the automation of these tools and disseminate best-practice techniques between software engineering and various scientific domains.

Probabilistic Computing Based on Random Devices

Recently, a new type of computing called as probabilistic computing approach has shown its broad application in various problems ranging from combinatorial optimizations and machine learning to quantum simulation where a randomly fluctuating bit (p-bit) constitutes a basic building block. This computing scheme tackles computationally hard problems in specific domains that can be efficiently solved using probabilistic algorithms compared to conventional deterministic computers. Here, we broadly review several aspects of this computing scheme starting from its building blocks and system network to working principles/algorithms and a family of problems that can be solved. By leveraging emerging nano-devices together with existing CMOS technology, we believe probabilistic computers can not only provide a valuable new approach to tackling computationally hard problems but also reveal the unknown power of probabilistic samplings by improving the speed and energy efficiency significantly.