Zheng Li

A Survey on Modeling Energy Consumption of Cloud Applications: Deconstruction, State of the Art, and Trade-Off Debates

2017 · IEEE Transactions on Sustainable Computing

Given the complexity and heterogeneity in Cloud computing scenarios, the modeling approach has widely been employed to investigate and analyze the energy consumption of Cloud applications, by abstracting real-world objects and processes that are difficult …

Research and design of recommendation in book circle system

2020 · Journal of Physics Conference Series

Abstract Reading is the ladder of human progress. It is the real needs of readers to quickly find good books and needed books in this fast-paced times. It will be more enjoyable if we can …

Hierarchical Attention Transfer Network for Cross-Domain Sentiment Classification

2018 · Proceedings of the AAAI Conference on Artificial Intelligence

Cross-domain sentiment classification aims to leverage useful information in a source domain to help do sentiment classification in a target domain that has no or little supervised information. Existing cross-domain sentiment classification methods cannot automatically …

Retrieval-Augmented Multilingual Keyphrase Generation with Retriever-Generator Iterative Training

2022 · Findings of the Association for Computational Linguistics: NAACL 2022

Keyphrase generation is the task of automatically predicting keyphrases given a piece of long text. Despite its recent flourishing, keyphrase generation on non-English languages haven't been vastly investigated. In this paper, we call attention to …

Robust Learning Against Label Noise Based on Activation Trend Tracking

2022 · IEEE Transactions on Instrumentation and Measurement

Defect detection of industrial products plays an important role in industrial production. Deep learning methods have been introduced to replace manual labor and save costs. Training deep neural networks requires large datasets with precisely annotated …

Backdoor Attacks Against Dataset Distillation

2023

Dataset distillation has emerged as a prominent technique to improve data efficiency when training machine learning models. It encapsulates the knowledge from a large dataset into a smaller synthetic dataset. A model trained on this …

NOTABLE: Transferable Backdoor Attacks Against Prompt-based NLP Models

2023 · arXiv (Cornell University)

Prompt-based learning is vulnerable to backdoor attacks. Existing backdoor attacks against prompt-based models consider injecting backdoors into the entire embedding layers or word embedding vectors. Such attacks can be easily affected by retraining on downstream …

FDINet: Protecting against DNN Model Extraction via Feature Distortion Index

2023 · arXiv (Cornell University)

Machine Learning as a Service (MLaaS) platforms have gained popularity due to their accessibility, cost-efficiency, scalability, and rapid development capabilities. However, recent research has highlighted the vulnerability of cloud-based models in MLaaS to model extraction …

Knowledge Graph Reasoning over Entities and Numerical Values

2023

A complex logic query in a knowledge graph refers to a query expressed in logic form that conveys a complex meaning, such as where did the Canadian Turing award winner graduate from? Knowledge graph reasoning-based …

Advancing Textual Prompt Learning with Anchored Attributes

2024 · arXiv (Cornell University)

Textual-based prompt learning methods primarily employ multiple learnable soft prompts and hard class tokens in a cascading manner as text inputs, aiming to align image and text (category) spaces for downstream tasks. However, current training …

Assessment of Large-Scale Solar Photovoltaic Potential on Building Roofs and Facades Using Geo-Aware Graph Attention Networks

2024

Assessing building photovoltaic (PV) potential is crucial for urban energy planning and achieving net-zero energy building goals. However, it still faces numerous limitations and challenges at the urban scale. Traditional solar irradiation simulation methods are …

A Deep Learning Based Assessment Method for Rehabilitation Exercises

2024

The most common symptom of stroke is limb dysfunction, resulting in a reduced quality of life for the patient. Patients can be rehabilitated through specific exercises movements. However, when the exercises are performed without the …

AndesVL Technical Report: An Efficient Mobile-side Multimodal Large Language Model

2025 · arXiv (Cornell University)

In recent years, while cloud-based MLLMs such as QwenVL, InternVL, GPT-4o, Gemini, and Claude Sonnet have demonstrated outstanding performance with enormous model sizes reaching hundreds of billions of parameters, they significantly surpass the limitations in …

Explainable Software Engineering: From State of the Practice to State of the Art

2025

The evidence-based paradigm has made software engineering (SE) more objective and less imprecise, facilitating extensive decision-making activities in software projects. However, by referring to the four-theory classificatory scheme, it can be observed that evidence-based SE …

DIAA: A Decoding-Efficient Inference Acceleration Approach for On-Device Large Language Models

2026 · Proceedings of the AAAI Conference on Artificial Intelligence

Large Language Models (LLMs) have revolutionized intelligent interactions, enabling mobile applications such as personal assistants on edge devices for local execution. Speculative decoding (SD) has emerged as a promising paradigm to accelerate LLM inference without …