InDUS: Incremental Document Understanding System Focus on Document Classification

d'Andecy, Vincent Poulain; Joseph, Aurélie; Ogier, Jean-Marc

doi:10.1109/das.2018.53

Cited by 8 publications

(4 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…INDUS [6], developed by the french company YOOZ, is design to classify document stream. This system relies on the A2ING classification algorithm, and uses the textual information as feature to characterize documents.…”

Section: Compared Methodsmentioning

confidence: 99%

“…However, two potential approaches have emerged recently: Structural Adaptation [25] or Training Adaptation [17] [28]. Out of neural network, several classifiers based on classical algorithm can be find like incremental SVM [20], K-means [1] and Growing Neural Gas (IGNG) [3] or industrial applications like INTELLIX [29] or InDUS [6].…”

Section: Overviewmentioning

confidence: 99%

“…This maintenance is costly in time and in workload because it needs to build a learning dataset. An option to overcome this problem is to use an incremental learning system like [29] or [6], but they are not able to compete with state of the art deep learning systems in performances.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Evaluation of Neural Network Classification Systems on Document Stream

Voerman¹,

Joseph²,

Coustaty³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

One major drawback of state of the art Neural Networks (NN)-based approaches for document classification purposes is the large number of training samples required to obtain an efficient classification. The minimum required number is around one thousand annotated documents for each class. In many cases it is very difficult, if not impossible, to gather this number of samples in real industrial processes. In this paper, we analyse the efficiency of NN-based document classification systems in a sub-optimal training case, based on the situation of a companys document stream. We evaluated three different approaches, one based on image content and two on textual content. The evaluation was divided into four parts: a reference case, to assess the performance of the system in the lab; two cases that each simulate a specific difficulty linked to document stream processing; and a realistic case that combined all of these difficulties. The realistic case highlighted the fact that there is a significant drop in the efficiency of NN-Based document classification systems. Although they remain efficient for well represented classes (with an over-fitting of the system for those classes), it is impossible for them to handle appropriately less well represented classes. NN-Based document classification systems need to be adapted to resolve these two problems before they can be considered for use in a companys document stream.

show abstract

Section: Compared Methodsmentioning

confidence: 99%

Section: Overviewmentioning

confidence: 99%

See 1 more Smart Citation

Evaluation of Neural Network Classification Systems on Document Stream

Voerman¹,

Joseph²,

Coustaty³

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Once the document is classified, an IE step is needed to extract relevant information for the Semantic Firewall, the goal is either to prevent the transfer of the whole document or just sensitive information in it.Many works have been done for IE from Image documents like in InforMys [22], Intellix [23], InDUS [24], CloudScan [25] or CUTIE [26]. The reader could find more details in those papers.…”

Section: Document Classification and Knowledge Extractionmentioning

confidence: 99%