Subtopic Deep Dive
Artificial Neural Networks
Research Guide
What is Artificial Neural Networks?
Artificial Neural Networks are computational models composed of interconnected nodes organized in layers, trained via backpropagation to approximate functions for tasks like classification and forecasting.
Research focuses on feedforward networks using backpropagation with optimization like conjugate gradient Fletcher-Reeves (Wanto et al., 2017, 109 citations) and activation functions such as sigmoid (Pratiwi et al., 2020, 152 citations). Convolutional Neural Networks (CNNs) enable image classification, as in melanoma detection (Refianti et al., 2019, 61 citations) and digit recognition (Peryanto et al., 2020, 69 citations). Over 1,000 papers explore variants for prediction and recognition since 2010.
Why It Matters
Neural networks power image classification for medical diagnostics, with CNNs classifying melanoma skin cancer (Refianti et al., 2019). Backpropagation predicts consumer price indices (Wanto et al., 2017) and student performance (Sumijan et al., 2016). They forecast population density (Wanto et al., 2017) and internet users (Setti and Wanto, 2019), enabling scalable pattern recognition across healthcare and economics.
Key Research Challenges
Activation Function Selection
Choosing optimal activation like sigmoid impacts model accuracy in backpropagation (Pratiwi et al., 2020). Different functions yield varying prediction errors in forecasting tasks. Researchers test sigmoid versus linear identity for density prediction (Wanto et al., 2017).
Backpropagation Optimization
Standard backpropagation converges slowly on large datasets, addressed by conjugate gradient Fletcher-Reeves (Wanto et al., 2017). Balancing learning rate and epochs remains critical for CPI prediction. Scalability to high-dimensional data persists as an issue (Siregar and Wanto, 2017).
Overfitting in Image Tasks
CNNs for handwritten digits and melanoma require regularization to avoid overfitting (Dixit et al., 2020; Refianti et al., 2019). K-fold cross-validation improves generalization (Peryanto et al., 2020). Limited training data exacerbates performance drops on unseen images.
Essential Papers
Comparative Analysis of KNN Algorithm using Various Normalization Techniques
Amit Pandey, Achin Jain · 2017 · International Journal of Computer Network and Information Security · 159 citations
Classification is the technique of identifying and assigning individual quantities to a group or a set.In pattern recognition, K-Nearest Neighbors algorithm is a non-parametric method for classific...
Sigmoid Activation Function in Selecting the Best Model of Artificial Neural Networks
Heny Pratiwi, Agus Perdana Windarto, S Susliansyah et al. · 2020 · Journal of Physics Conference Series · 152 citations
Abstract The aim of the research is to make predictions from the best architectural model of backpropagation neural networks. In determining the outcome in the form of a prediction model, the activ...
Use of Binary Sigmoid Function And Linear Identity In Artificial Neural Networks For Forecasting Population Density
Anjar Wanto, Agus Perdana Windarto, Dedy Hartama et al. · 2017 · IJISTECH (International Journal Of Information System & Technology) · 111 citations
Artificial Neural Network (ANN) is often used to solve forecasting cases. As in this study. The artificial neural network used is with backpropagation algorithm. The study focused on cases concerni...
Analysis of Artificial Neural Network Backpropagation Using Conjugate Gradient Fletcher Reeves In The Predicting Process
Anjar Wanto, Muhammad Zarlis, Sawaluddin Sawaluddin et al. · 2017 · Journal of Physics Conference Series · 109 citations
Backpropagation is a good artificial neural network algorithm used to predict, one of which is to predict the rate of Consumer Price Index (CPI) based on the foodstuff sector. While conjugate gradi...
Analysis of Artificial Neural Network Accuracy Using Backpropagation Algorithm In Predicting Process (Forecasting)
Sandy Putra Siregar, Anjar Wanto · 2017 · IJISTECH (International Journal Of Information System & Technology) · 102 citations
Artificial Neural Networks are a computational paradigm formed based on the neural structure of intelligent organisms to gain better knowledge. Artificial neural networks are often used for various...
Implementation of Neural Networks in Predicting the Understanding Level of Students Subject
Sumijan Sumijan, AgusPerdana Windarto, Abulwafa Muhammad et al. · 2016 · International Journal of Software Engineering and Its Applications · 82 citations
This paper implements artificial neuralnetworkin predictingthe understanding level ofstudent'scourse.By implementing artificial neural network based on backpropagation algorithm, an institution can...
Analysis of Backpropagation Algorithm in Predicting the Most Number of Internet Users in the World
Sunil Setti, Anjar Wanto · 2019 · Jurnal Online Informatika · 81 citations
The Internet today has become a primary need for its users. According to market research company e-Marketer, there are 25 countries with the largest internet users in the world. Indonesia is in the...
Reading Guide
Foundational Papers
Start with Narwan et al. (2014) for backpropagation in fault detection and Sureshkumar and Ravichandran (2010) for character recognition basics, as they introduce core training and segmentation techniques applied in modern nets.
Recent Advances
Study Pratiwi et al. (2020) for activation selection and Refianti et al. (2019) for CNN medical imaging, representing high-citation advances in optimization and vision.
Core Methods
Backpropagation algorithm with Fletcher-Reeves (Wanto et al., 2017), sigmoid/binary sigmoid activations (Pratiwi et al., 2020; Wanto et al., 2017), CNN with K-fold validation (Peryanto et al., 2020).
How PapersFlow Helps You Research Artificial Neural Networks
Discover & Search
Research Agent uses searchPapers with 'backpropagation conjugate gradient' to find Wanto et al. (2017, 109 citations), then citationGraph reveals 50+ related optimization papers and findSimilarPapers uncovers sigmoid variants like Pratiwi et al. (2020). exaSearch queries 'CNN melanoma classification' to surface Refianti et al. (2019).
Analyze & Verify
Analysis Agent applies readPaperContent to extract backpropagation hyperparameters from Wanto et al. (2017), verifies claims via verifyResponse (CoVe) against 10 similar papers, and uses runPythonAnalysis to reimplement sigmoid activation (Pratiwi et al., 2020) with NumPy for MSE computation. GRADE grading scores methodological rigor on 1-5 scale for forecasting accuracy.
Synthesize & Write
Synthesis Agent detects gaps in CNN regularization across Refianti et al. (2019) and Dixit et al. (2020), flags contradictions in activation choices. Writing Agent employs latexEditText for architecture diagrams, latexSyncCitations for 20-paper bibliography, and latexCompile for camera-ready review; exportMermaid generates network flowcharts.
Use Cases
"Reproduce backpropagation accuracy for CPI forecasting from Wanto et al. 2017"
Analysis Agent → readPaperContent (extracts hyperparameters) → runPythonAnalysis (NumPy backprop simulation with MSE=0.02) → researcher gets validated Python code and accuracy plot.
"Write LaTeX review of CNNs for melanoma classification citing Refianti 2019"
Synthesis Agent → gap detection (regularization needs) → Writing Agent → latexEditText (intro) → latexSyncCitations (10 papers) → latexCompile → researcher gets PDF with figures and citations.
"Find GitHub repos implementing conjugate gradient in neural nets"
Research Agent → searchPapers (Wanto 2017) → paperExtractUrls → paperFindGithubRepo → githubRepoInspect (3 repos) → researcher gets code diffs and training scripts.
Automated Workflows
Deep Research workflow scans 50+ backpropagation papers like Wanto et al. (2017), producing structured report with citation clusters via citationGraph. DeepScan applies 7-step CoVe to verify sigmoid claims (Pratiwi et al., 2020) with GRADE checkpoints. Theorizer generates hypotheses on hybrid activations from 20 forecasting papers.
Frequently Asked Questions
What defines Artificial Neural Networks?
Interconnected node layers trained by backpropagation for classification and forecasting, using activations like sigmoid.
What are common methods in this subtopic?
Backpropagation with conjugate gradient (Wanto et al., 2017), CNNs for images (Refianti et al., 2019), sigmoid activation optimization (Pratiwi et al., 2020).
What are key papers?
Pratiwi et al. (2020, 152 citations) on sigmoid; Wanto et al. (2017, 109 citations) on conjugate gradient; Refianti et al. (2019, 61 citations) on CNN melanoma classification.
What open problems exist?
Optimizing activations for large-scale forecasting, reducing CNN overfitting without massive data, scaling backpropagation beyond 150kV fault lines (Narwan et al., 2014).
Research Computer Science and Engineering with AI
PapersFlow provides specialized AI tools for Computer Science researchers. Here are the most relevant for this topic:
AI Literature Review
Automate paper discovery and synthesis across 474M+ papers
Code & Data Discovery
Find datasets, code repositories, and computational tools
Deep Research Reports
Multi-source evidence synthesis with counter-evidence
AI Academic Writing
Write research papers with AI assistance and LaTeX support
See how researchers in Computer Science & AI use PapersFlow
Field-specific workflows, example queries, and use cases.
Start Researching Artificial Neural Networks with AI
Search 474M+ papers, run AI-powered literature reviews, and write with integrated citations — all in one workspace.
See how PapersFlow works for Computer Science researchers
Part of the Computer Science and Engineering Research Guide