RecommenderSystems

🔅2025 Understanding Deep Learning 🌐 ّFree Download Full PDF 🔅 دانلود رایگان کتاب فهم یادگیری عمیق همراه با دسترسی به مثال ها و کدهای برنامه @book{prince2023understanding, author = "Simon J.D. Prince", title = "Understanding Deep Learning", publisher = "The MIT Press", year = 2023, url = "http://udlbook.com" } ▫️Coding exercises Python notebooks covering the whole text Sixty eight python notebook exercises with missing code to fill in based on the text #MIT #DL #DeepLearning #Deep_Learning #MIT_Press #Coding #Python @Recommender_Systems

2025 Enhanced multi-view graph convolutional networks for session-based recommendation 🌐 https://doi.org/10.1016/j.neucom.2025.130742 🌐 https://www.sciencedirect.com/science/article/abs/pii/S0925231225014146?via%3Dihub Existing session-based recommendation works have demonstrated significant advantages of graph neural networks in improving recommendation prediction capabilities. However, these models still have certain limitations when it comes to deeper handling of recommendation problems. (1) Previous models are relatively weak in capturing the complex higher-order information of item transition patterns, which to some extent affects the accuracy of recommendations. (2) Existing models mainly rely on item ID information to construct user preference models, overlooking the importance of revealing user preferences from a categorical perspective, thus limiting the degree of personalized recommendations. (3) Existing models fail to effectively capture consistency preferences between sessions, thereby affecting the coherence of recommendations and user experience. To overcome these limitations, we propose an innovative enhanced multi-view graph convolutional network model. Firstly, we introduce a hypergraph on top of the local and global graphs to further model high-order item transition relationships, fully utilizing complex associative information between items. Additionally, we design a fusion network to integrate different levels of item representations, enriching the expression of recommendation information. Secondly, we innovatively propose a category transition graph aimed at accurately capturing the transfer of user category interests within sessions, providing a more precise basis for personalized recommendations. Finally, we fully consider the similarity between sessions by constructing a session graph to extract consistency preferences, enhancing the coherence and accuracy of recommendation results. Extensive experiments conducted on two real-world public datasets demonstrate the outstanding performance of our approach in achieving state-of-the-art performance, providing new insights and solutions for the development of session-based recommendation field. #Enhanced #MultiView #Multi_View #Graph #Convolutional #NN #Neural_Networks #NeuralNetworks #Session #SessionBased #Session_Based #Personalized #Personalization @Recommender_Systems

2025 Spiking neural self-attention network for sequence recommendation 🌐 https://doi.org/10.1016/j.asoc.2024.112623 🌐 https://www.sciencedirect.com/science/article/abs/pii/S1568494624013978 Sequential recommendation plays an important role in providing a more personalized and accurate recommendation experience, and helps users discover new content that they may be interested in. However, sequence recommendation still faces long-term dependency issues (early behavior has a significant impact on subsequent recommendation results) and requiring high real-time performance. Moreover, the existing models based on recurrent networks or attention mechanisms still have shortcomings in addressing the long-term dependency problem and dynamic real-time performance of sequence recommendations. To address these challenges, we propose a long short-termmemory-spiking neural neural P (LSTM-SNP) self attention network, termed LSAF, for sequence recommendation that integrates long- and short-term user sequences. For this purpose, a three-channel structure is designed, where the long-term and short-term sequences are processed respectively by two self attention channels, and an LSTM-SNP channel is used to learn the user’s long-term dynamics. Then, these learned long-term and short-term features together are integrated by a self attention layer, and the prediction score and the predicted item (i.e. the next interaction item for user) can be obtained. In the LSAF model, the LSTM-SNP can effectively capture long-term and short-term dependency and nonlinear temporal characteristics. We evaluate the proposed LSAF model on three real-world datasets. The comparative experimental results with 13 baseline methods indicate that LSAF provides a competitive method for sequence recommendation. #Sequence #LSTM #LSTM_SNP #Nonlinear #Spiking #Neural #P #Sequential #Transformer #Nonlinear_Spiking_Neural_P_Systems #NonlinearSpiking #Neural_P_Systems #SelfAttention #Self_Attension #network @Recommender_Systems

2025 Adaptive in-context expert network with hierarchical data augmentation for sequential recommendation 🌐 https://doi.org/10.1016/j.knosys.2025.114061 🌐 https://www.sciencedirect.com/science/article/abs/pii/S0950705125011062 Transformer-based sequential recommendation systems have achieved remarkable success by modeling higher-order item dependencies. However, they often struggle to learn robust sequence representations in the presence of data sparsity, interaction noise, and cold-start issues. To address these issues, we propose a general framework named adaptive In-Context expert network with hierarchical Data augmentation for sequential Recommendation (CeDRec). Specifically, inspired by the hierarchical structure of text, we introduce a deep data augmentation method, DAug, which operates on higher-level semantic units to reconstruct users’ general preferences by capturing the interconnectivity of subsequences in latent space. Moreover, we design an adaptive in-context expert network that incorporates additional context experts to dynamically allocate attention weights, effectively alleviating the impact of noisy interactions. To further enhance the representation of unpopular items and improve overall recommendation quality, we introduce a mimic learning module that encourages popular items to approximate the distribution of unpopular ones, thereby providing richer self-supervision signals for underrepresented items. Extensive experiments show that CeDRec significantly outperforms state-of-the-art baselines in recommendation accuracy. Our model is capable of learning accurate and robust item representations even when faced with the problem of sparse interaction. #Adaptive #Context #Expert #Hierarchical #Augmentation #Sequential #Transformer #ColdStart #Cold_Start #Cold #CeDRec #Sparse #StateoftheArt @Recommender_Systems

🔅 آشنایی با REPLICATE ؛ کلکسیونی از ابزار های هوش مصنوعی...! ⬅️سرویس Replicate یه پلتفرم مشابه Hugging Face هست که به شما اجازه می‌ده تا مدل‌های هوش مصنوعی رو به راحتی اجرا کنید. این سایت به خصوص برای اجرای مدل‌هایی مثل استیبل دیفیوژن، KANDINSKY و مدل‌های دیگه فوق‌العاده‌ست. ⬅️تو سایت Replicate، می‌تونید به مدل‌های مختلف هوش مصنوعی که برای استفاده رایگان یا با هزینه کم در دسترس هستند، دسترسی پیدا کنید. برخی از مدل‌های معروف و محبوب شامل استیبل دفیوژن ؛ Dall-E ؛ GPT-4 ؛ KANDINSKY و... + با یک لاگین ساده می‌تونید به راحتی به مدل‌های مختلف دسترسی پیدا کنید و از قابلیت‌های اون‌ها بهره ببرید. https://replicate.com/ #AI #Research #Tools #ResearchTools #Research_Tools #AI_Tools #AiTools #Best #BestTools #Best_Tools #replicate #GPT @Recommender_Systems

🔅 برای افزایش کیفیت نگارش مقالات علمی، SciSpace پیشنهادهای زبانی و اصلاح ساختاری متن را ارائه می‌دهد و به کمک مدل‌های هوش مصنوعی، کمک می‌کند تا متون دقیق‌تر و حرفه‌ای‌تری تولید شوند. این پلتفرم به پژوهشگران و نویسندگان علمی کمک می‌کند تا با استفاده از ابزارهای پیشرفته، فرآیند نگارش مقالات و تحقیق را بهبود دهند. #AI #Research #Tools #ResearchTools #Research_Tools #SciSpace #Paper @Recommender_Systems

🔅 معرفی چند سایت کاربردی برای ویرایش متن مقالات : 1. Editage : یکی از بزرگ‌ترین و مشهورترین سایت‌های ویرایش مقالات با ارائه خدمات متنوع 2. Enago : سایتی برای ویرایش مقالات در تمامی زمینه‌های علمی با کارشناسانی مجرب و حرفه‌ای 3. Proof-Reading : یک سایت معتبر و جامع برای ویرایش مقالات، اسناد و متون تحقیقاتی 4. Scribendi : یک سایت ویرایش متون با تیمی از ویراستاران حرفه‌ای در بیش از ۴۰ زبان مختلف 5. Cambridge Proofreading LLC : یک سایت برای ویرایش متون تحقیقاتی و انتشاراتی با ارائه خدمات متفاوت و منحصر به فرد 6. Wordvice : یکی دیگر از سایت‌های مشهور و معتبر برای ویرایش مقالات و متون تحقیقاتی در زمینه‌های مختلف 7. ScienceDocs : یک سایت ویرایش مقالات و ترجمه متون تحقیقاتی در زمینه‌های مختلف از جمله علوم پزشکی، بیوشیمی و ریاضیات. #AI #Research #Tools #ResearchTools #Research_Tools #Explainpaper #Paper @Recommender_Systems

🔅 یک پلتفرم مبتنی بر هوش مصنوعی که هدف آن ساده‌سازی مفاهیم پیچیده مقالات علمی برای خوانندگان است: ⭕️ Explainpaper محققین می‌توانند فایل مقاله را آپلود کرده و با کلیک روی بخش‌های نامفهوم، توضیحاتی به زبان ساده و قابل‌فهم از هوش مصنوعی دریافت کنند #AI #Research #Tools #ResearchTools #Research_Tools #Explainpaper #Paper @Recommender_Systems