pith. sign in

arxiv: 2604.11932 · v1 · submitted 2026-04-13 · 💻 cs.CV

EigenCoin: sassanid coins classification based on Bhattacharyya distance

Rahele Allahverdi , Mohammad Mahdi Dehshibi , Azam Bastanfard , Daryoosh Akbarzadeh This is my paper

Pith reviewed 2026-05-10 16:26 UTC · model grok-4.3

classification 💻 cs.CV
keywords EigenCoinSassanid coinsBhattacharyya distancemanifold constructioncoin classificationimbalanced databasespattern recognitionoverfitting
0
0 comments X

The pith

EigenCoin classifies Sassanid coins from imbalanced data using manifold construction and Bhattacharyya distance, outperforming other algorithms with accuracies up to 21.75% while handling overfitting.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes EigenCoin to tackle pattern recognition on imbalanced databases by focusing on Sassanid coin images. It builds a manifold from the training coins, maps new test images into that space, and classifies them using Bhattacharyya distance. Experiments compare holistic and feature-based representations and report that the method beats the other algorithms tested, reaching accuracies between 9.45% and 21.75% while showing less overfitting. A sympathetic reader would care because many real-world image tasks, such as identifying historical objects, involve uneven numbers of examples per category, and a workable method here could ease similar recognition problems.

Core claim

EigenCoin consists of three main steps namely manifold construction, mapping test data, and classification. Conducted experiments show EigenCoin outperformed other observed algorithms and achieved the accuracy from 9.45% up to 21.75%, while it has the capability of handling the over-fitting problem. The focus includes testing the influence of the holistic and feature-based approaches.

What carries the argument

The EigenCoin manifold constructed from coin images, followed by mapping test data to the manifold and classification via Bhattacharyya distance, a statistical measure of divergence between distributions.

If this is right

  • The method provides higher accuracy than the compared algorithms for classifying Sassanid coins under data imbalance.
  • EigenCoin demonstrates a way to reduce overfitting in this type of pattern recognition task.
  • Holistic and feature-based approaches can be directly compared for effectiveness on coin images.
  • The three-step pipeline may transfer to other imbalanced image classification settings.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the manifold approach works here, the same steps could be tried on images of other ancient artifacts where some types appear far more often than others.
  • The modest accuracy numbers suggest that visual differences among Sassanid coins are subtle, possibly due to wear, lighting, or minting variations.
  • Testing the pipeline on datasets that explicitly report class counts and use cross-validation would show whether the overfitting resistance holds under stricter controls.

Load-bearing premise

The three-step process of building a manifold from the coin images, mapping new images into it, and classifying with Bhattacharyya distance reliably captures the variations between coin types and avoids overfitting even when accuracies stay low.

What would settle it

Re-running the method on a fresh collection of Sassanid coin images and obtaining accuracy no higher than the other algorithms, or clear signs of overfitting on validation data, would falsify the performance and generalization claims.

Figures

Figures reproduced from arXiv: 2604.11932 by Azam Bastanfard, Daryoosh Akbarzadeh, Mohammad Mahdi Dehshibi, Rahele Allahverdi.

Figure 1
Figure 1. Figure 1: Procedure of the proposed method 2.1. Extracting Region of Interest The contrast of the coin to be segmented differs greatly from the background, and gradient operators have the capability of detecting contrast changes in the image. Therefore, edge information from Sobel operator is utilized to obtain a binary mask which facilitates the segmentation process. The binary gradient mask highlights lines of hig… view at source ↗
read the original abstract

Solving pattern recognition problems using imbalanced databases is a hot topic, which entices researchers to bring it into focus. Therefore, we consider this problem in the application of Sassanid coins classification. Our focus is not only on proposing EigenCoin manifold with Bhattacharyya distance for the classification task, but also on testing the influence of the holistic and feature-based approaches. EigenCoin consists of three main steps namely manifold construction, mapping test data, and classification. Conducted experiments show EigenCoin outperformed other observed algorithms and achieved the accuracy from 9.45% up to 21.75%, while it has the capability of handling the over-fitting problem.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 1 minor

Summary. The paper proposes EigenCoin, a three-step manifold-based approach (manifold construction, test data mapping, and Bhattacharyya distance classification) for classifying Sassanid coins from imbalanced datasets. It claims that EigenCoin outperforms unspecified other algorithms, achieves accuracies between 9.45% and 21.75%, and handles overfitting, while also examining holistic versus feature-based approaches.

Significance. If the experimental claims can be substantiated with proper baselines and dataset statistics, the work could offer a niche contribution to pattern recognition for cultural heritage artifacts with class imbalance. However, the reported absolute accuracies are low, and without context on the number of classes or baseline performance, it is unclear whether the method advances the state of the art or merely addresses a difficult multi-class problem.

major comments (3)
  1. [Abstract] Abstract: The central claim that 'EigenCoin outperformed other observed algorithms' cannot be assessed because the manuscript provides neither the identities of the baseline algorithms nor their accuracy scores for comparison.
  2. [Abstract] Abstract: The reported accuracy range (9.45%–21.75%) is impossible to interpret as success or outperformance without the number of classes N, total sample count, per-class distribution, or train/test protocol; random-guess performance would be 1/N, yet none of these quantities are stated.
  3. [Abstract] Abstract: No equations, pseudocode, or derivation steps are supplied for the EigenCoin manifold construction or the Bhattacharyya-distance classifier, preventing verification that the three-step process is well-defined and does not reduce to a fitted or circular procedure.
minor comments (1)
  1. [Abstract] Abstract: The phrasing 'testing the influence of the holistic and feature-based approaches' is ambiguous; it is unclear whether these are alternatives to EigenCoin or components within it.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We have reviewed the comments carefully and will make revisions to improve the clarity of the abstract and the methodological details. Our point-by-point responses to the major comments follow.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that 'EigenCoin outperformed other observed algorithms' cannot be assessed because the manuscript provides neither the identities of the baseline algorithms nor their accuracy scores for comparison.

    Authors: We agree that the abstract does not identify the baseline algorithms or report their accuracy scores, which makes the outperformance claim difficult to evaluate. In the revised manuscript we will update the abstract to explicitly name the algorithms used in the comparisons and include their accuracy scores alongside the EigenCoin results. revision: yes

  2. Referee: [Abstract] Abstract: The reported accuracy range (9.45%–21.75%) is impossible to interpret as success or outperformance without the number of classes N, total sample count, per-class distribution, or train/test protocol; random-guess performance would be 1/N, yet none of these quantities are stated.

    Authors: We agree that the abstract requires additional context on the dataset and protocol to allow readers to interpret the accuracy figures meaningfully. We will revise the abstract to include the number of classes, total sample count, per-class distribution summary, and the train/test protocol used. revision: yes

  3. Referee: [Abstract] Abstract: No equations, pseudocode, or derivation steps are supplied for the EigenCoin manifold construction or the Bhattacharyya-distance classifier, preventing verification that the three-step process is well-defined and does not reduce to a fitted or circular procedure.

    Authors: We acknowledge that the manuscript does not supply equations, pseudocode, or derivation steps for the manifold construction and Bhattacharyya-distance classifier. In the revised version we will add the relevant equations, derivation steps, and pseudocode to the methods section so that the three-step process can be fully verified. revision: yes

Circularity Check

0 steps flagged

No derivation chain or equations present; claims are purely empirical.

full rationale

The provided abstract and description outline EigenCoin as a three-step procedural method (manifold construction, test data mapping, Bhattacharyya classification) with experimental accuracy results (9.45%-21.75%) and an overfitting claim. No mathematical derivations, equations, first-principles predictions, fitted parameters, or self-citations appear that could reduce to inputs by construction. Performance statements rest on conducted experiments rather than any analytic chain, so no circularity patterns (self-definitional, fitted-input-as-prediction, etc.) can be identified. The paper is self-contained as a high-level method proposal plus empirical report.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 1 invented entities

Abstract-only review; no explicit free parameters, axioms, or invented entities are detailed beyond naming the EigenCoin manifold. The method implicitly relies on standard assumptions of manifold learning and distance metrics without stating them.

invented entities (1)
  • EigenCoin manifold no independent evidence
    purpose: To represent and classify Sassanid coin images in a reduced space
    Introduced as the central contribution but lacks independent evidence or falsifiable predictions outside the reported experiments.

pith-pipeline@v0.9.0 · 5420 in / 1286 out tokens · 82337 ms · 2026-05-10T16:26:05.848632+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

7 extracted references · 7 canonical work pages

  1. [1]

    [1]. M. Kampel, M. Zaharieva, Recognizing Ancient Coins Based on Local Features, Lecture Notes in Computer Science, Vol. 5358, 2008, pp. 11-22. Class Method Hormozd IV Khosrow I Khosrow II Hormozd V Overall Harris (184D) 37.93% 12.5% 74.83% 0% 18.82% EigenCoin (112D) 24.1% 31.2% 75.5% 0% 21.75% 1160 R. Allahverdi et al./ AWERProcedia Information Technolog...

    work page 2008

  2. [2]

    [4]. A. Khashman, B. Sekeroglu, K. Dimililer, Intelligent Coin Identification System, in: Proceedings of the 21th IEEE International Symposio m on Intelligent Control, 2006, pp. 1226-1230. [5]. L. Van Der Maaten, P. Poon, Coin-o-matic: A fast system for reliable coin classification, in: Muscle CIS Coin Competition Workshop, 2006, pp. 7-18. [6]. W. Zuo, D....

    work page 2006

  3. [3]

    [7]. C. Garcia, G. Zikos, G. Tziritas, A wavelet-based framework for face recognition, in: Proceedings of European Conference on Computer Vision, 2008, pp. 1-

    work page 2008

  4. [4]

    [8]. C. Harris, M. Stephens, A combined corner and edge detector, in: Proceedings of the 4th Alvey Vision Conference, 1988, pp. 147-

    work page 1988

  5. [5]

    [10]. M. Turk, A. Pentland, Eigenfaces for recognition, Journal of Cognitive Neuroscience 3(1) (1991) 71 –86. [11]. S. M. Smith, J. M. Brady, SUSAN - a new approach to low level image processing, International Journal of Computer Vision 23 (1) (1997) 45-78. [12]. P.V.C. Hough, Machine Analysis of Bubble Chamber Pictures, in: Proceedings of Internati onal ...

    work page 1991

  6. [6]

    [13]. D. G. Lowe, Object recognition from local scale-invariant features, in: Proceedings of thr 7th IEEE International Conference on Computer Vision,1999, pp. 1150-

    work page 1999

  7. [7]

    [14]. H. Bay, T. Tuytelaars, L. Van Gool, SURF: Speeded Up Robust Features, Computer Vision and Image Understanding (CVIU) 110(3) (2008) 346-359. [15]. K. Mikolajczyk, C. Schmid, A performance evaluation of local descriptors, IEEE Transactions on Pattern An alysis and Machine Intelligence 10(27) (2005) 1615-

    work page 2008