# Seminar-Materials **Repository Path**: amtech/Seminar-Materials ## Basic Information - **Project Name**: Seminar-Materials - **Description**: 组会ppt与论文--每一次的精心准备都值得留下记录 - **Primary Language**: Unknown - **License**: Not specified - **Default Branch**: master - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 0 - **Forks**: 2 - **Created**: 2024-01-21 - **Last Updated**: 2024-01-21 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README # 数据科学研讨会记录（信息系统组） ### 在线教育相关paper | Title | Detail | Author | link | | ------------------------------------------------------------ | ------------------------------------------------ | ------ | ------------------------------------------------------------ | | Prerequisite Relation Learning for Concepts in MOOCs（ACL2017） | 研究MOOC中概念之间的先后关系 | Fuyingnan | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/Education/fuyingnan.pptx) | | Question Difficulty Prediction for READING Problems in Standard Tests (AAAI2017) | 标准测试需要保持每次测试的难度相近，因此对测试问题的难度估计十分关键，而传统的难度估计需要大量的专家人力成本。因此本文提出一个新型的TACNN（Test-aware Attention-based Convolutional Neural Network）网络架构解决了标准测试中阅读理解题的难题预测问题。真实数据集的扩展实验验证了TACNN的有效性。 | Zhurenyu | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/Education/zhurenyu1.pptx) | | Identifying At-Risk Students in Massive Open Online Courses (AAAI2015) | 对MOOC平台处于border line的学生进行及时激励 | Hanyi | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/Education/hanyi.pptx) | | Investigating Active Learning for Concept Prerequisite Learning（AAAI2018） | 研究对比 了各种主动学习方法在学科先决关系预测中的性能好坏 | Chenyuanzhe | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/Education/Concept_prerequisite_learning.pptx) | | Predicting Instructor’s Intervention in MOOC forums（ACL2014） | 在MOOC的论坛中，预测某个帖子是否需要指导老师干预 | Lina | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/Education/PrecidctingInstructorsInterventioninMOOCforums.pdf) | | MOOC Dropout Prediction: Lessons Learned from Making Pipelines Interpretable（WWW17） | 探讨MOOC课程平台Dropout预测模型的可解释性 | KuangJun | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/Education/InterpretableDropout.pptx) | | Mining MOOC Clickstreams: On the Relationship Between Learner Behavior and Performance (KDD2015) | 在MOOC平台上，发现学习者反复出现的行为，分析行为和表现的关系。 | Liutingting | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/Education/liutingting.pptx) | | Understanding Dropouts in MOOCs (AAAI) | 探讨MOOC平台注册用户的动机，辍学的原因，以及如何提前预测辍学以便人工干预 | Gaobaoli | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/Education/gaobaoli.pdf) | | Dropout Model Evaluation in MOOCs (AAAI2018) | 对于预测MOOC上的学生是否辍学的模型，用 Friedman 检验和Nemenyi 后续检验 ，从算法（比如CART，ada）和特征类别（forum-,assignment-, and clickstream-based ）两方面结合，在统计的意义上评估了它们的性能区别。 | Zhouxiaoxu | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/Education/zhouxiaoxu.pptx) | | Recovering Concept Prerequisite Relations from University Course Dependencies (AAAI2017) | 构建带权路径和发现知识的先后关系 | Zhengshu | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/Education/zhengshu.pptx) | ### ### 2018-11-09 | Title | Detail | Author | link | | ---------------------------------------- | ---------------------------------------- | --------- | --------- | | Improving Knowledge Graph Embedding Using Simple Constraints (ACL2018) | 将KG embedding为复数的形式，通过新增复数运算的约束来提高KG embedding的效果| Zhouxiaoxu | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/2018-2019/zhouxiaoxu.pptx) | ### 2018-11-02 | Title | Detail | Author | link | | ---------------------------------------- | ---------------------------------------- | --------- | --------- | | Knowledge-aware Attentive Neural Network for Ranking Question Answer Pairs (SIGIR2018) | 主要做的是问题答案的排序，利用知识图谱扩展句子的表示学习。| Liutingting | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/2018-2019/liutingting.pptx) | ### 2018-10-27 | Title | Detail | Author | link | | ---------------------------------------- | ---------------------------------------- | --------- | --------- | | CoLink: An Unsupervised Framework for User Identity Linkage (AAAI2018) | 一个无监督的框架，用于进行不同网络之间的实体匹配(即找到结点之间的映射关系) | Lina | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/2018-2019/lina.pptx) | ### 2018-10-20 | Title | Detail | Author | link | | ---------------------------------------- | ---------------------------------------- | --------- | --------- | | A Walk-based Model on Entity Graphs for Relation Extraction (ACL2018) | 进行关系抽取时，考虑了任意两两实体对间的关系构成的网络 | Kuangjun | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/2018-2019/kuangjun.pptx) | ### 2018-10-13 | Title | Detail | Author | link | | ---------------------------------------- | ---------------------------------------- | --------- | --------- | | Contextual String Embeddings for Sequence Labeling (COLING2018) | （flair）预训练一个字符级别的语言模型，用其来对单词进行表征。通过将其与传统的word embedding进行拼接，提高了序列标注任务的性能 | Chenyuanzhe | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/2018-2019/flair.pptx) | ### 2018-9-28 | Title | Detail | Author | link | | ---------------------------------------- | ---------------------------------------- | --------- | --------- | | Question Difficulty Prediction for READING Problems in Standard Tests (AAAI2017) | 构造了TACNN框架来评估英语阅读问题难度，将文章、问题、选项作为特征输入，预测每个问题的难度系数 | Zhurenyu | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/2018-2019/AAAI%202017%20Question%20Difficulty%20Prediction%20for%20READING%20Problems%20in%20Standard%20Tests.pptx) | ### 2018-9-21 | Title | Detail | Author | link | | ---------------------------------------- | ---------------------------------------- | --------- | --------- | | Learning from Semi-Supervised Weak-Label Data(AAAI2018) | 半监督弱标签下的多标签分类算法(数据集只有少部分标注，大部分未标注；数据标注的准确性无法保证) | Fuyingnan | [ppt](https://github.com/ECNUdase/Seminar-Materials/blob/master/2018-2019/Learning%20from%20Semi-Supervised%20Weak-Label%20Data.pptx?raw=true) | ### 2018-3-16 | Title | Detail | Author | | ---------------------------------------- | ---------------------------------------- | --------- | | Why should I trust you | 提出了开源工具"Lime"，能够解释样本的预测结果，并且增加模型本身的可解释性 | Fuyingnan | | Deep Residual Learning for Image Recognition | 提出了更深层次的卷积网络架构——残差网络，解决了传统模型中网络难以训练的问题 | Zhurenyu | | A Unified Probabilistic Framework for Name Disambiguation in Digital Library | 将姓名消歧问题formalize成一个隐马尔科夫随机场，并提出了参数估计的两阶段算法；提出了自动确定重名人数的auto K算法 | Lina | | JointExtractionofEntitiesandRelations | 将实体识别和关系提取统一为序列标注问题，使用同一个模型同时进行实体识别和关系提取 | Kuangjun | ### 2018-3-23 | Title | Detail | Author | | ---------------------------------------- | -------------------------- | ----------- | | Mask R-CNN | 提出了Mask R-CNN用于图像的实例分割 | yuruonan | | Deep Reinforcement Learning for Mention-Ranking | 采用神经网络和强化学习技术增加共指消解的准确率 | chenyuanzhe | | Question Answering with Subgraph Embeddings | 采用基于子图嵌入的方法，进行问答系统的训练和答案预测 | tanglumin | | RNN学习心得 | 介绍了RNN相关概念，讲解了梯度消失和权重冲突问题 | yangkang | ### 2018-3-30 | Title | Detail | Author | | ---------------------------------------- | ---------------------------------------- | ---------- | | Reinforcement Learning for Relation Classification from Noisy Data | 提出一个新的关系分类模型，由实体选择器和关系分类器构成，能够在“Sentence Level”提取关系。将实体选择问题转换成强化学习问题。 | GuHang | | Pix2code: Generating Code from a Graphical User Interface Screenshot | 使用CNN和RNN的联合模型，将网页的UI图转化为对应的HTML代码 | E Shen | | JAVA GC机制 | 讲解了java的内存分配机制和垃圾回收机制 | YinJiaLing | ### 2018-4-13 | Title | Detail | Author | | ---------------------------------------- | ---------------------------------------- | ---------- | | Convolutional Sequence to Sequence Learning | An architecture based entirely on convolutional neural networks for sequence to sequence learning(such as NMT) | CuiYiFeng | | DeepFM：A Factorization-Machine based Neural Network for CTR Predicti | 回顾了过去的CTR模型，以及介绍了一系列基于深度学习的CTR模型（FNN,PNN,WDL） | ChenLeiHui | | Aspect Level Sentiment Classification with Deep Memory Network | 介绍了Memory Network，用于情感分析问题 | Void-Yu | ### 2018-4-20 | Title | Detail | Author | | ---------------------------------------- | ---------------------------------------- | --------- | | Learning Structured Representation for Text Classification via Reinforcement Learning | 使用ID-LSTM + HS-LSTM学习文本结构，并用策略梯度法进行强化学习 | JinLiJiao | | Human Action Adverb Recognition: ADHA Dataset And A Three-stream Hybrid Model | 贡献了一个数据集：x为人类动作的视频流序列，y为动作对应的副词。 例如识别接吻的视频是“甜蜜地”，"激动地"，“绅士地” ... | SunChen | ### 2018-5-4 | Title | Detail | Author | | ---------------------------------------- | --------------------------------------- | --------- | | Deep Forest: Towards An Alternative to Deep Neural Network | 周志华提出的gcForest多粒度级联森林 | FuYingNan | | Structure Regularized Neural Network for Entity Relation Classification for Chinese Literature Text | 利用结构正则化简化句法结构，进行关系提取 | KuangJun | | Ranking-Based Name Matching for Author Disambiguation in Bibliographic Data | KDD Cup 2013第二名，使用基于字符串和元路径的相似度进行作者姓名消歧 | LiNa | ### 2018-5-18 | Title | Detail | Author | | ---------------------------------------- | ---------------------------------------- | ----------- | | Modeling Mention, Context and Entity with Neural Networks for Entity Disambiguation | 2015年实体消岐的最优模型，采用神经网络，使用了Embedding，卷积，神经张量网络等结构。 | ChenYuanZhe | | 解析HashTable，HashMap，ConcurrentHashMap | 讨论了java中该三种结构的特点，主要从多线程安全性、性能等方面考虑 | YinJiaLing | ### 2018-6-8 | Title | Detail | Author | | ---------------------------------------- | ---------------------------------------- | ---------- | | Study about word embedding on sentiment subspace | 研究词向量中用于表达情感的向量子空间，目的是提高情感分类任务效果 | Void-Yu | ### 2018-6-15 | Title | Detail | Author | | ---------------------------------------- | ---------------------------------------- | ---------- | | R-FCN: Object Detection via Region-based Fully Convolutional Networks | 目标检测网络。效果不比之前的RCNN，Fast-RCNN差，但是速度更快了。 | God E | | BiNE: Bipartite Network Embedding | 在二分图中采用了表示学习的方法，将节点embedding成向量，通过向量距离来度量节点的相似性。 训练过程类似Word2vec，使用了负采样，负样本的采样分布使用了LSH来代替频率 | ChenLeiHui |