Mahmood Amintoosi; Eisa Kohan-Baghkheirati
Abstract
Every year, extensive experimental analysis is conducted to evaluate the anti-cancer properties of plants. Developing a well-ranked list of potential anti-cancer plants based on verified anti-cancer metabolites can significantly reduce the time and cost required for plant evaluation. ...
Read More
Every year, extensive experimental analysis is conducted to evaluate the anti-cancer properties of plants. Developing a well-ranked list of potential anti-cancer plants based on verified anti-cancer metabolites can significantly reduce the time and cost required for plant evaluation. This paper proposes a method for generating such a ranked list by analyzing biological graphs of plant-metabolite interactions. In this approach, graph nodes are ranked based on specific graph features. However, a challenge arises in selecting the most informative graph features that ensure the resulting ranked plant list is more relevant, prioritizing plants with greater anti-cancer properties at the top. To address this challenge, we propose the use of the Average Precision metric commonly used in information retrieval and recommender systems, to compare different ranked lists. By constructing a network that captures the similarities between plants based on their shared metabolites, and ranking plants using different combinations of graph features, we can identify the subset of features that yields a ranked list with a higher Average Precision score. This subset of features can then be considered the most suitable for recommending anti-cancer plants. The proposed method can be used to select the best graph features for screening unverified plant lists for anti-cancer properties, increasing the likelihood of identifying plants with higher scores in the list that possess anti-cancer properties.
Hadis Ahmadian Yazdi; Seyed Javad Seyyed Mahdavi Chabok; Maryam KheirAbadi
Abstract
In recent decades, the amount and variety of data have grown rapidly. As a result, data storage, compression, and analysis have become critical subjects in data mining and machine learning. It is essential to achieve accurate compression without losing important data in the process. Therefore, this ...
Read More
In recent decades, the amount and variety of data have grown rapidly. As a result, data storage, compression, and analysis have become critical subjects in data mining and machine learning. It is essential to achieve accurate compression without losing important data in the process. Therefore, this work proposes an effective data compression method for recommender systems based on the attention mechanism. The proposed method performs data compression on two levels: features and records. It is time-aware and based on time windows, taking into account users' activity and preventing the loss of important data. The resulting technique can be efficiently utilized for deep networks, where the amount of data is a significant challenge. Experimental results demonstrate that this technique not only reduces the amount of data and processing time but also achieves acceptable accuracy.
