![](/client/images/blank.gif)
Eylem Seç
![CR-GAN: Content-Based Recommender System with Conditional Generative Adversarial Networks için kapak resmi CR-GAN: Content-Based Recommender System with Conditional Generative Adversarial Networks için kapak resmi](/client/assets/d79c3e4af2b6d196/ctx/images/no_image.png)
CR-GAN: Content-Based Recommender System with Conditional Generative Adversarial Networks
Başlık:
CR-GAN: Content-Based Recommender System with Conditional Generative Adversarial Networks
Yazar:
Vundela, Karthik Reddy, author.
ISBN:
9780438075382
Yazar Ek Girişi:
Fiziksel Tanımlama:
1 electronic resource (95 pages)
Genel Not:
Source: Masters Abstracts International, Volume: 57-06M(E).
Advisors: Yugyung Lee Committee members: Vijay Kumar; Yongjie Zheng.
Özet:
Recommender systems have become increasingly popular by providing a wide range of products with a variety of styles. This trend has resulted in consumers expecting more intelligent and highly dynamic recommenders. The traditional recommender systems mainly rely on the historical or static data like user choices and ratings. Even the machine learning-based existing recommender systems are mainly based on pre-trained models or in the absence of visual cues in the corpus. With the ever-growing volume of visual data as well as revolutionary advances in deep learning (DL) for computer vision, incorporating the visual aspects of the data (e.g., color, fashion styles) yields more relevant suggestions in contrast to the traditional approaches.
In this thesis, we propose the CR-GAN recommender, a new type of recommender system through the integration of content-based filtering and deep learning (DL). The CR-GAN recommender aims to generate the design of new products according to the styles suggested by the recommender that has a capability for generation of the design of products using conditional Generative Adversarial Networks (GAN). The CR-GAN recommender is different from the traditional recommender systems that are conducting recommendations by retrieving existing products from the corpus. The CR-GAN recommender recommends newly generated clothes to a user based on the visual aspects (color and style) of the input image. We achieve this through the following process: 1) Feed an input image to CNN to recognize different attributes like category and color of the clothes, 2) Make the recommendations according to the CNN classification using Content-based similarity filtering and co-occurrence, 3) Generate new clothes by using the attributes of the recommendations as conditions for the Conditional Generative Adversarial Network (GAN). The CR-GAN recommender has been evaluated by taking into consideration the performance of individual models (CNN Classifier, Recommender, GAN). The outcomes of the networks and their performance at different stages were also validated.
Notlar:
School code: 0134
Konu Başlığı:
Tüzel Kişi Ek Girişi:
Mevcut:*
Yer Numarası | Demirbaş Numarası | Shelf Location | Lokasyon / Statüsü / İade Tarihi |
---|---|---|---|
XX(694801.1) | 694801-1001 | Proquest E-Tez Koleksiyonu | Arıyor... |
On Order
Liste seç
Bunu varsayılan liste yap.
Öğeler başarıyla eklendi
Öğeler eklenirken hata oldu. Lütfen tekrar deneyiniz.
:
Select An Item
Data usage warning: You will receive one text message for each title you selected.
Standard text messaging rates apply.