A5073519838- Recommender Systems and Techniques
- Advanced Bandit Algorithms Research
- Video Analysis and Summarization
- Image and Video Quality Assessment
- Data Stream Mining Techniques
- Tracheal and airway disorders
- Congenital Diaphragmatic Hernia Studies
- Head and Neck Surgical Oncology
- Topic Modeling
- Industrial Vision Systems and Defect Detection
- Wikis in Education and Collaboration
- Digital Rights Management and Security
- Viral Infections and Immunology Research
- Advanced Graph Neural Networks
- Semantic Web and Ontologies
- Multimedia Communication and Technology
- Advanced Vision and Imaging
- Esophageal and GI Pathology
- Image Retrieval and Classification Techniques
- Media Influence and Politics
- Congenital Heart Disease Studies
- Advanced Causal Inference Techniques
- Stochastic Gradient Optimization Techniques
- Image Processing Techniques and Applications
- Sentiment Analysis and Opinion Mining
Google (United States)
2023-2024
Shandong University
2023
Qilu Hospital of Shandong University
2022
Effective exploration is believed to positively influence the long-term user experience on recommendation platforms. Determining its exact benefits, however, has been challenging. Regular A/B tests often measure neutral or even negative engagement metrics while failing capture benefits. We here introduce new experiment designs formally quantify value of by examining effects content corpus, and connecting corpus growth from real-world experiments. Once established values exploration, we...
Recommender systems are essential for finding personalized content users on online platforms. These often trained historical user interaction data, which collects feedback system recommendations. This creates a loop leading to popularity bias; popular is over-represented in the better learned, and thus recommended even more. Less struggles reach its potential audiences. Popularity bias limits diversity of that exposed to, makes it harder new creators gain traction. Existing methods alleviate...
Exposure bias and its induced feedback loop effect are well-known problems in recommender systems. Exploration is believed to be the key break such loops. While classical contextual bandit algorithms as Upper-Confidence-Bound Thompson Sampling have been successful addressing exploration-exploitation trade-off single-task settings with one clear reward signal, modern systems often leverage multiple rich sources of clicks, likes, dislikes, shares, satisfaction survey responses, employ...
Sequential recommenders have been widely used in industry due to their strength modeling user preferences. While these models excel at learning a user's positive interests, less attention has paid from negative feedback. Negative feedback is an important lever of control, and comes with expectation that should respond quickly reduce similar recommendations the user. However, signals are often ignored training objective sequential retrieval models, which primarily aim predicting interactions....
In recent years, social media users spend significant amount of time on Short-Form Video (SFV) platforms. Its success in creating an immersive viewership experience is not only from the content, but also due to its unique UI innovation: instead providing choices for click, SFV platforms actively recommend content watch one at a time. this paper, we highlight challenges rooted such changes recommendation system design. Firstly, there yet much unexplored sources biases under new UI, as are no...
Fine-tuning is becoming widely used for leveraging the power of pre-trained foundation models in new downstream tasks. While there are many successes fine-tuning on various tasks, recent studies have observed challenges generalization fine-tuned to unseen distributions (i.e., out-of-distribution; OOD). To improve OOD generalization, some previous identify limitations data and regulate preserve general representation learned from pre-training data. However, potential often ignored. In this...
Traditional recommendation systems are subject to a strong feedback loop by learning from and reinforcing past user-item interactions, which in turn limits the discovery of novel user interests. To address this, we introduce hybrid hierarchical framework combining Large Language Models (LLMs) classic models for interest exploration. The controls interfacing between LLMs through "interest clusters", granularity can be explicitly determined algorithm designers. It recommends next interests...
Traditional recommendation systems are subject to a strong feedback loop by learning from and reinforcing past user-item interactions, which in turn limits the discovery of novel user interests. To address this, we introduce hybrid hierarchical framework combining Large Language Models (LLMs) classic models for interest exploration. The controls interfacing between LLMs through "interest clusters", granularity can be explicitly determined algorithm designers. It recommends next interests...
Sequential models are invaluable for powering personalized recommendation systems. In the context of short-form video (SFV) feeds, where user behavior history is typically longer, systems must be able to understand users' long-term interests. However, deploying large sequence extensive web-scale applications faces challenges due high serving cost. To address this, we propose an industrial framework designed efficiently models. Specifically, proposed infrastructure decouples model and main...
The insider threat is increasingly becoming extremely important for companies, organizations and even governments. A malicious, or a careless, can cause severe damage to the resources reputation of an organization. In this article, ...
Multi-task prediction models and value are the de-facto standard ranking components in modern large-scale content recommendation systems. However, they typically optimized to model users' passive consumption behaviors, rank a way grow only consumption-centric values. In this talk, we discuss key insight that it is possible sparse participatory content-generation actions as well ecosystem through new system. We made following technical contributions system: (1) introducing for generation...
Effective exploration is believed to positively influence the long-term user experience on recommendation platforms. Determining its exact benefits, however, has been challenging. Regular A/B tests often measure neutral or even negative engagement metrics while failing capture benefits. We here introduce new experiment designs formally quantify value of by examining effects content corpus, and connecting corpus growth from real-world experiments. Once established values exploration, we...
Objective: To examine the effect of surgical treatment in children with pulmonary artery sling and strategy. Methods: Relevant data 110 admitted to Department Cardiac Surgery, Children's Hospital Affiliated Shandong University from February 2017 July 2022 were retrospectively analyzed. There 55 males females, aging (M(IQR)) 9.0 (10.6) months (range: 1 96 months). The weight was 7.8 (3.5) kg 2.5 25.0 kg). Of patients, 108 had different degrees tracheal stenosis 2 normal trachea. Left...
Objective: To examine the outcomes of Slide tracheoplasty for children with severe congenital tracheal stenosis received previous repeated balloon dilatation or metal stent placement under endoscopy. Methods: A retrospective study was conducted in 9 undergoing interventional therapy tracheoscopy and later due to obvious respiratory symptoms at Department Cardiac Surgery, Qilu Children's Hospital Shandong University between February 2017 July 2021. There were 7 males 2 females a median age...