Temporal knowledge representation and reasoning techniques using time Petri nets

2022年考研考博-考博英语-浙江大学考试全真模拟易错、难点剖析AB卷（带答案）一.综合题(共15题)1.单选题It is astonishing to know that children and youth ( ) the biggest segment of the country's homeless population.问题1选项A.substitute forB.make outC.make upD.make up for【答案】C【解析】substitute for代替，取代；make out理解，认出；make up有很多意思，这里指组成；make up for弥补，补偿。

句意：令人惊讶的是，儿童和青年构成了该国无家可归人口的最大部分。

选项C符合句意。

2.单选题A theory is an organized set of principles that is designed to explain and predict some phenomenon. Good theories also provide specific testable predictions, or ( ) about the relation between two or more variables.问题1选项A.hypothesisB.conceptionC.ideaD.meaning【答案】A【解析】hypothesis假设；conception概念，设想；idea想法，主意；meaning意义，意图。

句意：理论是一套有组织的原理，用来解释和预测一些现象。

好的理论也提供了具体的可测试的预测，或者关于两个或两个以上变量之间关系的假设。

选项A符合句意。

3.单选题It is disturbing to note how many crimes we do know about were detected（） , not by systematic inspections or other security procedures.问题1选项A.by accidentB.on scheduleC.in generalD.at intervals【答案】A【解析】by accident偶然，意外的；on schedule按时；in general总之，通常；at intervals不时。

机器学习与人工智能领域中常用的英语词汇1.General Concepts (基础概念)•Artificial Intelligence (AI) - 人工智能1)Artificial Intelligence (AI) - 人工智能2)Machine Learning (ML) - 机器学习3)Deep Learning (DL) - 深度学习4)Neural Network - 神经网络5)Natural Language Processing (NLP) - 自然语言处理6)Computer Vision - 计算机视觉7)Robotics - 机器人技术8)Speech Recognition - 语音识别9)Expert Systems - 专家系统10)Knowledge Representation - 知识表示11)Pattern Recognition - 模式识别12)Cognitive Computing - 认知计算13)Autonomous Systems - 自主系统14)Human-Machine Interaction - 人机交互15)Intelligent Agents - 智能代理16)Machine Translation - 机器翻译17)Swarm Intelligence - 群体智能18)Genetic Algorithms - 遗传算法19)Fuzzy Logic - 模糊逻辑20)Reinforcement Learning - 强化学习•Machine Learning (ML) - 机器学习1)Machine Learning (ML) - 机器学习2)Artificial Neural Network - 人工神经网络3)Deep Learning - 深度学习4)Supervised Learning - 有监督学习5)Unsupervised Learning - 无监督学习6)Reinforcement Learning - 强化学习7)Semi-Supervised Learning - 半监督学习8)Training Data - 训练数据9)Test Data - 测试数据10)Validation Data - 验证数据11)Feature - 特征12)Label - 标签13)Model - 模型14)Algorithm - 算法15)Regression - 回归16)Classification - 分类17)Clustering - 聚类18)Dimensionality Reduction - 降维19)Overfitting - 过拟合20)Underfitting - 欠拟合•Deep Learning (DL) - 深度学习1)Deep Learning - 深度学习2)Neural Network - 神经网络3)Artificial Neural Network (ANN) - 人工神经网络4)Convolutional Neural Network (CNN) - 卷积神经网络5)Recurrent Neural Network (RNN) - 循环神经网络6)Long Short-Term Memory (LSTM) - 长短期记忆网络7)Gated Recurrent Unit (GRU) - 门控循环单元8)Autoencoder - 自编码器9)Generative Adversarial Network (GAN) - 生成对抗网络10)Transfer Learning - 迁移学习11)Pre-trained Model - 预训练模型12)Fine-tuning - 微调13)Feature Extraction - 特征提取14)Activation Function - 激活函数15)Loss Function - 损失函数16)Gradient Descent - 梯度下降17)Backpropagation - 反向传播18)Epoch - 训练周期19)Batch Size - 批量大小20)Dropout - 丢弃法•Neural Network - 神经网络1)Neural Network - 神经网络2)Artificial Neural Network (ANN) - 人工神经网络3)Deep Neural Network (DNN) - 深度神经网络4)Convolutional Neural Network (CNN) - 卷积神经网络5)Recurrent Neural Network (RNN) - 循环神经网络6)Long Short-Term Memory (LSTM) - 长短期记忆网络7)Gated Recurrent Unit (GRU) - 门控循环单元8)Feedforward Neural Network - 前馈神经网络9)Multi-layer Perceptron (MLP) - 多层感知器10)Radial Basis Function Network (RBFN) - 径向基函数网络11)Hopfield Network - 霍普菲尔德网络12)Boltzmann Machine - 玻尔兹曼机13)Autoencoder - 自编码器14)Spiking Neural Network (SNN) - 脉冲神经网络15)Self-organizing Map (SOM) - 自组织映射16)Restricted Boltzmann Machine (RBM) - 受限玻尔兹曼机17)Hebbian Learning - 海比安学习18)Competitive Learning - 竞争学习19)Neuroevolutionary - 神经进化20)Neuron - 神经元•Algorithm - 算法1)Algorithm - 算法2)Supervised Learning Algorithm - 有监督学习算法3)Unsupervised Learning Algorithm - 无监督学习算法4)Reinforcement Learning Algorithm - 强化学习算法5)Classification Algorithm - 分类算法6)Regression Algorithm - 回归算法7)Clustering Algorithm - 聚类算法8)Dimensionality Reduction Algorithm - 降维算法9)Decision Tree Algorithm - 决策树算法10)Random Forest Algorithm - 随机森林算法11)Support Vector Machine (SVM) Algorithm - 支持向量机算法12)K-Nearest Neighbors (KNN) Algorithm - K近邻算法13)Naive Bayes Algorithm - 朴素贝叶斯算法14)Gradient Descent Algorithm - 梯度下降算法15)Genetic Algorithm - 遗传算法16)Neural Network Algorithm - 神经网络算法17)Deep Learning Algorithm - 深度学习算法18)Ensemble Learning Algorithm - 集成学习算法19)Reinforcement Learning Algorithm - 强化学习算法20)Metaheuristic Algorithm - 元启发式算法•Model - 模型1)Model - 模型2)Machine Learning Model - 机器学习模型3)Artificial Intelligence Model - 人工智能模型4)Predictive Model - 预测模型5)Classification Model - 分类模型6)Regression Model - 回归模型7)Generative Model - 生成模型8)Discriminative Model - 判别模型9)Probabilistic Model - 概率模型10)Statistical Model - 统计模型11)Neural Network Model - 神经网络模型12)Deep Learning Model - 深度学习模型13)Ensemble Model - 集成模型14)Reinforcement Learning Model - 强化学习模型15)Support Vector Machine (SVM) Model - 支持向量机模型16)Decision Tree Model - 决策树模型17)Random Forest Model - 随机森林模型18)Naive Bayes Model - 朴素贝叶斯模型19)Autoencoder Model - 自编码器模型20)Convolutional Neural Network (CNN) Model - 卷积神经网络模型•Dataset - 数据集1)Dataset - 数据集2)Training Dataset - 训练数据集3)Test Dataset - 测试数据集4)Validation Dataset - 验证数据集5)Balanced Dataset - 平衡数据集6)Imbalanced Dataset - 不平衡数据集7)Synthetic Dataset - 合成数据集8)Benchmark Dataset - 基准数据集9)Open Dataset - 开放数据集10)Labeled Dataset - 标记数据集11)Unlabeled Dataset - 未标记数据集12)Semi-Supervised Dataset - 半监督数据集13)Multiclass Dataset - 多分类数据集14)Feature Set - 特征集15)Data Augmentation - 数据增强16)Data Preprocessing - 数据预处理17)Missing Data - 缺失数据18)Outlier Detection - 异常值检测19)Data Imputation - 数据插补20)Metadata - 元数据•Training - 训练1)Training - 训练2)Training Data - 训练数据3)Training Phase - 训练阶段4)Training Set - 训练集5)Training Examples - 训练样本6)Training Instance - 训练实例7)Training Algorithm - 训练算法8)Training Model - 训练模型9)Training Process - 训练过程10)Training Loss - 训练损失11)Training Epoch - 训练周期12)Training Batch - 训练批次13)Online Training - 在线训练14)Offline Training - 离线训练15)Continuous Training - 连续训练16)Transfer Learning - 迁移学习17)Fine-Tuning - 微调18)Curriculum Learning - 课程学习19)Self-Supervised Learning - 自监督学习20)Active Learning - 主动学习•Testing - 测试1)Testing - 测试2)Test Data - 测试数据3)Test Set - 测试集4)Test Examples - 测试样本5)Test Instance - 测试实例6)Test Phase - 测试阶段7)Test Accuracy - 测试准确率8)Test Loss - 测试损失9)Test Error - 测试错误10)Test Metrics - 测试指标11)Test Suite - 测试套件12)Test Case - 测试用例13)Test Coverage - 测试覆盖率14)Cross-Validation - 交叉验证15)Holdout Validation - 留出验证16)K-Fold Cross-Validation - K折交叉验证17)Stratified Cross-Validation - 分层交叉验证18)Test Driven Development (TDD) - 测试驱动开发19)A/B Testing - A/B 测试20)Model Evaluation - 模型评估•Validation - 验证1)Validation - 验证2)Validation Data - 验证数据3)Validation Set - 验证集4)Validation Examples - 验证样本5)Validation Instance - 验证实例6)Validation Phase - 验证阶段7)Validation Accuracy - 验证准确率8)Validation Loss - 验证损失9)Validation Error - 验证错误10)Validation Metrics - 验证指标11)Cross-Validation - 交叉验证12)Holdout Validation - 留出验证13)K-Fold Cross-Validation - K折交叉验证14)Stratified Cross-Validation - 分层交叉验证15)Leave-One-Out Cross-Validation - 留一法交叉验证16)Validation Curve - 验证曲线17)Hyperparameter Validation - 超参数验证18)Model Validation - 模型验证19)Early Stopping - 提前停止20)Validation Strategy - 验证策略•Supervised Learning - 有监督学习1)Supervised Learning - 有监督学习2)Label - 标签3)Feature - 特征4)Target - 目标5)Training Labels - 训练标签6)Training Features - 训练特征7)Training Targets - 训练目标8)Training Examples - 训练样本9)Training Instance - 训练实例10)Regression - 回归11)Classification - 分类12)Predictor - 预测器13)Regression Model - 回归模型14)Classifier - 分类器15)Decision Tree - 决策树16)Support Vector Machine (SVM) - 支持向量机17)Neural Network - 神经网络18)Feature Engineering - 特征工程19)Model Evaluation - 模型评估20)Overfitting - 过拟合21)Underfitting - 欠拟合22)Bias-Variance Tradeoff - 偏差-方差权衡•Unsupervised Learning - 无监督学习1)Unsupervised Learning - 无监督学习2)Clustering - 聚类3)Dimensionality Reduction - 降维4)Anomaly Detection - 异常检测5)Association Rule Learning - 关联规则学习6)Feature Extraction - 特征提取7)Feature Selection - 特征选择8)K-Means - K均值9)Hierarchical Clustering - 层次聚类10)Density-Based Clustering - 基于密度的聚类11)Principal Component Analysis (PCA) - 主成分分析12)Independent Component Analysis (ICA) - 独立成分分析13)T-distributed Stochastic Neighbor Embedding (t-SNE) - t分布随机邻居嵌入14)Gaussian Mixture Model (GMM) - 高斯混合模型15)Self-Organizing Maps (SOM) - 自组织映射16)Autoencoder - 自动编码器17)Latent Variable - 潜变量18)Data Preprocessing - 数据预处理19)Outlier Detection - 异常值检测20)Clustering Algorithm - 聚类算法•Reinforcement Learning - 强化学习1)Reinforcement Learning - 强化学习2)Agent - 代理3)Environment - 环境4)State - 状态5)Action - 动作6)Reward - 奖励7)Policy - 策略8)Value Function - 值函数9)Q-Learning - Q学习10)Deep Q-Network (DQN) - 深度Q网络11)Policy Gradient - 策略梯度12)Actor-Critic - 演员-评论家13)Exploration - 探索14)Exploitation - 开发15)Temporal Difference (TD) - 时间差分16)Markov Decision Process (MDP) - 马尔可夫决策过程17)State-Action-Reward-State-Action (SARSA) - 状态-动作-奖励-状态-动作18)Policy Iteration - 策略迭代19)Value Iteration - 值迭代20)Monte Carlo Methods - 蒙特卡洛方法•Semi-Supervised Learning - 半监督学习1)Semi-Supervised Learning - 半监督学习2)Labeled Data - 有标签数据3)Unlabeled Data - 无标签数据4)Label Propagation - 标签传播5)Self-Training - 自训练6)Co-Training - 协同训练7)Transudative Learning - 传导学习8)Inductive Learning - 归纳学习9)Manifold Regularization - 流形正则化10)Graph-based Methods - 基于图的方法11)Cluster Assumption - 聚类假设12)Low-Density Separation - 低密度分离13)Semi-Supervised Support Vector Machines (S3VM) - 半监督支持向量机14)Expectation-Maximization (EM) - 期望最大化15)Co-EM - 协同期望最大化16)Entropy-Regularized EM - 熵正则化EM17)Mean Teacher - 平均教师18)Virtual Adversarial Training - 虚拟对抗训练19)Tri-training - 三重训练20)Mix Match - 混合匹配•Feature - 特征1)Feature - 特征2)Feature Engineering - 特征工程3)Feature Extraction - 特征提取4)Feature Selection - 特征选择5)Input Features - 输入特征6)Output Features - 输出特征7)Feature Vector - 特征向量8)Feature Space - 特征空间9)Feature Representation - 特征表示10)Feature Transformation - 特征转换11)Feature Importance - 特征重要性12)Feature Scaling - 特征缩放13)Feature Normalization - 特征归一化14)Feature Encoding - 特征编码15)Feature Fusion - 特征融合16)Feature Dimensionality Reduction - 特征维度减少17)Continuous Feature - 连续特征18)Categorical Feature - 分类特征19)Nominal Feature - 名义特征20)Ordinal Feature - 有序特征•Label - 标签1)Label - 标签2)Labeling - 标注3)Ground Truth - 地面真值4)Class Label - 类别标签5)Target Variable - 目标变量6)Labeling Scheme - 标注方案7)Multi-class Labeling - 多类别标注8)Binary Labeling - 二分类标注9)Label Noise - 标签噪声10)Labeling Error - 标注错误11)Label Propagation - 标签传播12)Unlabeled Data - 无标签数据13)Labeled Data - 有标签数据14)Semi-supervised Learning - 半监督学习15)Active Learning - 主动学习16)Weakly Supervised Learning - 弱监督学习17)Noisy Label Learning - 噪声标签学习18)Self-training - 自训练19)Crowdsourcing Labeling - 众包标注20)Label Smoothing - 标签平滑化•Prediction - 预测1)Prediction - 预测2)Forecasting - 预测3)Regression - 回归4)Classification - 分类5)Time Series Prediction - 时间序列预测6)Forecast Accuracy - 预测准确性7)Predictive Modeling - 预测建模8)Predictive Analytics - 预测分析9)Forecasting Method - 预测方法10)Predictive Performance - 预测性能11)Predictive Power - 预测能力12)Prediction Error - 预测误差13)Prediction Interval - 预测区间14)Prediction Model - 预测模型15)Predictive Uncertainty - 预测不确定性16)Forecast Horizon - 预测时间跨度17)Predictive Maintenance - 预测性维护18)Predictive Policing - 预测式警务19)Predictive Healthcare - 预测性医疗20)Predictive Maintenance - 预测性维护•Classification - 分类1)Classification - 分类2)Classifier - 分类器3)Class - 类别4)Classify - 对数据进行分类5)Class Label - 类别标签6)Binary Classification - 二元分类7)Multiclass Classification - 多类分类8)Class Probability - 类别概率9)Decision Boundary - 决策边界10)Decision Tree - 决策树11)Support Vector Machine (SVM) - 支持向量机12)K-Nearest Neighbors (KNN) - K最近邻算法13)Naive Bayes - 朴素贝叶斯14)Logistic Regression - 逻辑回归15)Random Forest - 随机森林16)Neural Network - 神经网络17)SoftMax Function - SoftMax函数18)One-vs-All (One-vs-Rest) - 一对多(一对剩余)19)Ensemble Learning - 集成学习20)Confusion Matrix - 混淆矩阵•Regression - 回归1)Regression Analysis - 回归分析2)Linear Regression - 线性回归3)Multiple Regression - 多元回归4)Polynomial Regression - 多项式回归5)Logistic Regression - 逻辑回归6)Ridge Regression - 岭回归7)Lasso Regression - Lasso回归8)Elastic Net Regression - 弹性网络回归9)Regression Coefficients - 回归系数10)Residuals - 残差11)Ordinary Least Squares (OLS) - 普通最小二乘法12)Ridge Regression Coefficient - 岭回归系数13)Lasso Regression Coefficient - Lasso回归系数14)Elastic Net Regression Coefficient - 弹性网络回归系数15)Regression Line - 回归线16)Prediction Error - 预测误差17)Regression Model - 回归模型18)Nonlinear Regression - 非线性回归19)Generalized Linear Models (GLM) - 广义线性模型20)Coefficient of Determination (R-squared) - 决定系数21)F-test - F检验22)Homoscedasticity - 同方差性23)Heteroscedasticity - 异方差性24)Autocorrelation - 自相关25)Multicollinearity - 多重共线性26)Outliers - 异常值27)Cross-validation - 交叉验证28)Feature Selection - 特征选择29)Feature Engineering - 特征工程30)Regularization - 正则化2.Neural Networks and Deep Learning (神经网络与深度学习)•Convolutional Neural Network (CNN) - 卷积神经网络1)Convolutional Neural Network (CNN) - 卷积神经网络2)Convolution Layer - 卷积层3)Feature Map - 特征图4)Convolution Operation - 卷积操作5)Stride - 步幅6)Padding - 填充7)Pooling Layer - 池化层8)Max Pooling - 最大池化9)Average Pooling - 平均池化10)Fully Connected Layer - 全连接层11)Activation Function - 激活函数12)Rectified Linear Unit (ReLU) - 线性修正单元13)Dropout - 随机失活14)Batch Normalization - 批量归一化15)Transfer Learning - 迁移学习16)Fine-Tuning - 微调17)Image Classification - 图像分类18)Object Detection - 物体检测19)Semantic Segmentation - 语义分割20)Instance Segmentation - 实例分割21)Generative Adversarial Network (GAN) - 生成对抗网络22)Image Generation - 图像生成23)Style Transfer - 风格迁移24)Convolutional Autoencoder - 卷积自编码器25)Recurrent Neural Network (RNN) - 循环神经网络•Recurrent Neural Network (RNN) - 循环神经网络1)Recurrent Neural Network (RNN) - 循环神经网络2)Long Short-Term Memory (LSTM) - 长短期记忆网络3)Gated Recurrent Unit (GRU) - 门控循环单元4)Sequence Modeling - 序列建模5)Time Series Prediction - 时间序列预测6)Natural Language Processing (NLP) - 自然语言处理7)Text Generation - 文本生成8)Sentiment Analysis - 情感分析9)Named Entity Recognition (NER) - 命名实体识别10)Part-of-Speech Tagging (POS Tagging) - 词性标注11)Sequence-to-Sequence (Seq2Seq) - 序列到序列12)Attention Mechanism - 注意力机制13)Encoder-Decoder Architecture - 编码器-解码器架构14)Bidirectional RNN - 双向循环神经网络15)Teacher Forcing - 强制教师法16)Backpropagation Through Time (BPTT) - 通过时间的反向传播17)Vanishing Gradient Problem - 梯度消失问题18)Exploding Gradient Problem - 梯度爆炸问题19)Language Modeling - 语言建模20)Speech Recognition - 语音识别•Long Short-Term Memory (LSTM) - 长短期记忆网络1)Long Short-Term Memory (LSTM) - 长短期记忆网络2)Cell State - 细胞状态3)Hidden State - 隐藏状态4)Forget Gate - 遗忘门5)Input Gate - 输入门6)Output Gate - 输出门7)Peephole Connections - 窥视孔连接8)Gated Recurrent Unit (GRU) - 门控循环单元9)Vanishing Gradient Problem - 梯度消失问题10)Exploding Gradient Problem - 梯度爆炸问题11)Sequence Modeling - 序列建模12)Time Series Prediction - 时间序列预测13)Natural Language Processing (NLP) - 自然语言处理14)Text Generation - 文本生成15)Sentiment Analysis - 情感分析16)Named Entity Recognition (NER) - 命名实体识别17)Part-of-Speech Tagging (POS Tagging) - 词性标注18)Attention Mechanism - 注意力机制19)Encoder-Decoder Architecture - 编码器-解码器架构20)Bidirectional LSTM - 双向长短期记忆网络•Attention Mechanism - 注意力机制1)Attention Mechanism - 注意力机制2)Self-Attention - 自注意力3)Multi-Head Attention - 多头注意力4)Transformer - 变换器5)Query - 查询6)Key - 键7)Value - 值8)Query-Value Attention - 查询-值注意力9)Dot-Product Attention - 点积注意力10)Scaled Dot-Product Attention - 缩放点积注意力11)Additive Attention - 加性注意力12)Context Vector - 上下文向量13)Attention Score - 注意力分数14)SoftMax Function - SoftMax函数15)Attention Weight - 注意力权重16)Global Attention - 全局注意力17)Local Attention - 局部注意力18)Positional Encoding - 位置编码19)Encoder-Decoder Attention - 编码器-解码器注意力20)Cross-Modal Attention - 跨模态注意力•Generative Adversarial Network (GAN) - 生成对抗网络1)Generative Adversarial Network (GAN) - 生成对抗网络2)Generator - 生成器3)Discriminator - 判别器4)Adversarial Training - 对抗训练5)Minimax Game - 极小极大博弈6)Nash Equilibrium - 纳什均衡7)Mode Collapse - 模式崩溃8)Training Stability - 训练稳定性9)Loss Function - 损失函数10)Discriminative Loss - 判别损失11)Generative Loss - 生成损失12)Wasserstein GAN (WGAN) - Wasserstein GAN（WGAN）13)Deep Convolutional GAN (DCGAN) - 深度卷积生成对抗网络（DCGAN）14)Conditional GAN (c GAN) - 条件生成对抗网络（c GAN）15)Style GAN - 风格生成对抗网络16)Cycle GAN - 循环生成对抗网络17)Progressive Growing GAN (PGGAN) - 渐进式增长生成对抗网络（PGGAN）18)Self-Attention GAN (SAGAN) - 自注意力生成对抗网络（SAGAN）19)Big GAN - 大规模生成对抗网络20)Adversarial Examples - 对抗样本•Encoder-Decoder - 编码器-解码器1)Encoder-Decoder Architecture - 编码器-解码器架构2)Encoder - 编码器3)Decoder - 解码器4)Sequence-to-Sequence Model (Seq2Seq) - 序列到序列模型5)State Vector - 状态向量6)Context Vector - 上下文向量7)Hidden State - 隐藏状态8)Attention Mechanism - 注意力机制9)Teacher Forcing - 强制教师法10)Beam Search - 束搜索11)Recurrent Neural Network (RNN) - 循环神经网络12)Long Short-Term Memory (LSTM) - 长短期记忆网络13)Gated Recurrent Unit (GRU) - 门控循环单元14)Bidirectional Encoder - 双向编码器15)Greedy Decoding - 贪婪解码16)Masking - 遮盖17)Dropout - 随机失活18)Embedding Layer - 嵌入层19)Cross-Entropy Loss - 交叉熵损失20)Tokenization - 令牌化•Transfer Learning - 迁移学习1)Transfer Learning - 迁移学习2)Source Domain - 源领域3)Target Domain - 目标领域4)Fine-Tuning - 微调5)Domain Adaptation - 领域自适应6)Pre-Trained Model - 预训练模型7)Feature Extraction - 特征提取8)Knowledge Transfer - 知识迁移9)Unsupervised Domain Adaptation - 无监督领域自适应10)Semi-Supervised Domain Adaptation - 半监督领域自适应11)Multi-Task Learning - 多任务学习12)Data Augmentation - 数据增强13)Task Transfer - 任务迁移14)Model Agnostic Meta-Learning (MAML) - 与模型无关的元学习（MAML）15)One-Shot Learning - 单样本学习16)Zero-Shot Learning - 零样本学习17)Few-Shot Learning - 少样本学习18)Knowledge Distillation - 知识蒸馏19)Representation Learning - 表征学习20)Adversarial Transfer Learning - 对抗迁移学习•Pre-trained Models - 预训练模型1)Pre-trained Model - 预训练模型2)Transfer Learning - 迁移学习3)Fine-Tuning - 微调4)Knowledge Transfer - 知识迁移5)Domain Adaptation - 领域自适应6)Feature Extraction - 特征提取7)Representation Learning - 表征学习8)Language Model - 语言模型9)Bidirectional Encoder Representations from Transformers (BERT) - 双向编码器结构转换器10)Generative Pre-trained Transformer (GPT) - 生成式预训练转换器11)Transformer-based Models - 基于转换器的模型12)Masked Language Model (MLM) - 掩蔽语言模型13)Cloze Task - 填空任务14)Tokenization - 令牌化15)Word Embeddings - 词嵌入16)Sentence Embeddings - 句子嵌入17)Contextual Embeddings - 上下文嵌入18)Self-Supervised Learning - 自监督学习19)Large-Scale Pre-trained Models - 大规模预训练模型•Loss Function - 损失函数1)Loss Function - 损失函数2)Mean Squared Error (MSE) - 均方误差3)Mean Absolute Error (MAE) - 平均绝对误差4)Cross-Entropy Loss - 交叉熵损失5)Binary Cross-Entropy Loss - 二元交叉熵损失6)Categorical Cross-Entropy Loss - 分类交叉熵损失7)Hinge Loss - 合页损失8)Huber Loss - Huber损失9)Wasserstein Distance - Wasserstein距离10)Triplet Loss - 三元组损失11)Contrastive Loss - 对比损失12)Dice Loss - Dice损失13)Focal Loss - 焦点损失14)GAN Loss - GAN损失15)Adversarial Loss - 对抗损失16)L1 Loss - L1损失17)L2 Loss - L2损失18)Huber Loss - Huber损失19)Quantile Loss - 分位数损失•Activation Function - 激活函数1)Activation Function - 激活函数2)Sigmoid Function - Sigmoid函数3)Hyperbolic Tangent Function (Tanh) - 双曲正切函数4)Rectified Linear Unit (Re LU) - 矩形线性单元5)Parametric Re LU (P Re LU) - 参数化Re LU6)Exponential Linear Unit (ELU) - 指数线性单元7)Swish Function - Swish函数8)Softplus Function - Soft plus函数9)Softmax Function - SoftMax函数10)Hard Tanh Function - 硬双曲正切函数11)Softsign Function - Softsign函数12)GELU (Gaussian Error Linear Unit) - GELU（高斯误差线性单元）13)Mish Function - Mish函数14)CELU (Continuous Exponential Linear Unit) - CELU（连续指数线性单元）15)Bent Identity Function - 弯曲恒等函数16)Gaussian Error Linear Units (GELUs) - 高斯误差线性单元17)Adaptive Piecewise Linear (APL) - 自适应分段线性函数18)Radial Basis Function (RBF) - 径向基函数•Backpropagation - 反向传播1)Backpropagation - 反向传播2)Gradient Descent - 梯度下降3)Partial Derivative - 偏导数4)Chain Rule - 链式法则5)Forward Pass - 前向传播6)Backward Pass - 反向传播7)Computational Graph - 计算图8)Neural Network - 神经网络9)Loss Function - 损失函数10)Gradient Calculation - 梯度计算11)Weight Update - 权重更新12)Activation Function - 激活函数13)Optimizer - 优化器14)Learning Rate - 学习率15)Mini-Batch Gradient Descent - 小批量梯度下降16)Stochastic Gradient Descent (SGD) - 随机梯度下降17)Batch Gradient Descent - 批量梯度下降18)Momentum - 动量19)Adam Optimizer - Adam优化器20)Learning Rate Decay - 学习率衰减•Gradient Descent - 梯度下降1)Gradient Descent - 梯度下降2)Stochastic Gradient Descent (SGD) - 随机梯度下降3)Mini-Batch Gradient Descent - 小批量梯度下降4)Batch Gradient Descent - 批量梯度下降5)Learning Rate - 学习率6)Momentum - 动量7)Adaptive Moment Estimation (Adam) - 自适应矩估计8)RMSprop - 均方根传播9)Learning Rate Schedule - 学习率调度10)Convergence - 收敛11)Divergence - 发散12)Adagrad - 自适应学习速率方法13)Adadelta - 自适应增量学习率方法14)Adamax - 自适应矩估计的扩展版本15)Nadam - Nesterov Accelerated Adaptive Moment Estimation16)Learning Rate Decay - 学习率衰减17)Step Size - 步长18)Conjugate Gradient Descent - 共轭梯度下降19)Line Search - 线搜索20)Newton's Method - 牛顿法•Learning Rate - 学习率1)Learning Rate - 学习率2)Adaptive Learning Rate - 自适应学习率3)Learning Rate Decay - 学习率衰减4)Initial Learning Rate - 初始学习率5)Step Size - 步长6)Momentum - 动量7)Exponential Decay - 指数衰减8)Annealing - 退火9)Cyclical Learning Rate - 循环学习率10)Learning Rate Schedule - 学习率调度11)Warm-up - 预热12)Learning Rate Policy - 学习率策略13)Learning Rate Annealing - 学习率退火14)Cosine Annealing - 余弦退火15)Gradient Clipping - 梯度裁剪16)Adapting Learning Rate - 适应学习率17)Learning Rate Multiplier - 学习率倍增器18)Learning Rate Reduction - 学习率降低19)Learning Rate Update - 学习率更新20)Scheduled Learning Rate - 定期学习率•Batch Size - 批量大小1)Batch Size - 批量大小2)Mini-Batch - 小批量3)Batch Gradient Descent - 批量梯度下降4)Stochastic Gradient Descent (SGD) - 随机梯度下降5)Mini-Batch Gradient Descent - 小批量梯度下降6)Online Learning - 在线学习7)Full-Batch - 全批量8)Data Batch - 数据批次9)Training Batch - 训练批次10)Batch Normalization - 批量归一化11)Batch-wise Optimization - 批量优化12)Batch Processing - 批量处理13)Batch Sampling - 批量采样14)Adaptive Batch Size - 自适应批量大小15)Batch Splitting - 批量分割16)Dynamic Batch Size - 动态批量大小17)Fixed Batch Size - 固定批量大小18)Batch-wise Inference - 批量推理19)Batch-wise Training - 批量训练20)Batch Shuffling - 批量洗牌•Epoch - 训练周期1)Training Epoch - 训练周期2)Epoch Size - 周期大小3)Early Stopping - 提前停止4)Validation Set - 验证集5)Training Set - 训练集6)Test Set - 测试集7)Overfitting - 过拟合8)Underfitting - 欠拟合9)Model Evaluation - 模型评估10)Model Selection - 模型选择11)Hyperparameter Tuning - 超参数调优12)Cross-Validation - 交叉验证13)K-fold Cross-Validation - K折交叉验证14)Stratified Cross-Validation - 分层交叉验证15)Leave-One-Out Cross-Validation (LOOCV) - 留一法交叉验证16)Grid Search - 网格搜索17)Random Search - 随机搜索18)Model Complexity - 模型复杂度19)Learning Curve - 学习曲线20)Convergence - 收敛3.Machine Learning Techniques and Algorithms (机器学习技术与算法)•Decision Tree - 决策树1)Decision Tree - 决策树2)Node - 节点3)Root Node - 根节点4)Leaf Node - 叶节点5)Internal Node - 内部节点6)Splitting Criterion - 分裂准则7)Gini Impurity - 基尼不纯度8)Entropy - 熵9)Information Gain - 信息增益10)Gain Ratio - 增益率11)Pruning - 剪枝12)Recursive Partitioning - 递归分割13)CART (Classification and Regression Trees) - 分类回归树14)ID3 (Iterative Dichotomiser 3) - 迭代二叉树315)C4.5 (successor of ID3) - C4.5（ID3的后继者）16)C5.0 (successor of C4.5) - C5.0（C4.5的后继者）17)Split Point - 分裂点18)Decision Boundary - 决策边界19)Pruned Tree - 剪枝后的树20)Decision Tree Ensemble - 决策树集成•Random Forest - 随机森林1)Random Forest - 随机森林2)Ensemble Learning - 集成学习3)Bootstrap Sampling - 自助采样4)Bagging (Bootstrap Aggregating) - 装袋法5)Out-of-Bag (OOB) Error - 袋外误差6)Feature Subset - 特征子集7)Decision Tree - 决策树8)Base Estimator - 基础估计器9)Tree Depth - 树深度10)Randomization - 随机化11)Majority Voting - 多数投票12)Feature Importance - 特征重要性13)OOB Score - 袋外得分14)Forest Size - 森林大小15)Max Features - 最大特征数16)Min Samples Split - 最小分裂样本数17)Min Samples Leaf - 最小叶节点样本数18)Gini Impurity - 基尼不纯度19)Entropy - 熵20)Variable Importance - 变量重要性•Support Vector Machine (SVM) - 支持向量机1)Support Vector Machine (SVM) - 支持向量机2)Hyperplane - 超平面3)Kernel Trick - 核技巧4)Kernel Function - 核函数5)Margin - 间隔6)Support Vectors - 支持向量7)Decision Boundary - 决策边界8)Maximum Margin Classifier - 最大间隔分类器9)Soft Margin Classifier - 软间隔分类器10) C Parameter - C参数11)Radial Basis Function (RBF) Kernel - 径向基函数核12)Polynomial Kernel - 多项式核13)Linear Kernel - 线性核14)Quadratic Kernel - 二次核15)Gaussian Kernel - 高斯核16)Regularization - 正则化17)Dual Problem - 对偶问题18)Primal Problem - 原始问题19)Kernelized SVM - 核化支持向量机20)Multiclass SVM - 多类支持向量机•K-Nearest Neighbors (KNN) - K-最近邻1)K-Nearest Neighbors (KNN) - K-最近邻2)Nearest Neighbor - 最近邻3)Distance Metric - 距离度量4)Euclidean Distance - 欧氏距离5)Manhattan Distance - 曼哈顿距离6)Minkowski Distance - 闵可夫斯基距离7)Cosine Similarity - 余弦相似度8)K Value - K值9)Majority Voting - 多数投票10)Weighted KNN - 加权KNN11)Radius Neighbors - 半径邻居12)Ball Tree - 球树13)KD Tree - KD树14)Locality-Sensitive Hashing (LSH) - 局部敏感哈希15)Curse of Dimensionality - 维度灾难16)Class Label - 类标签17)Training Set - 训练集18)Test Set - 测试集19)Validation Set - 验证集20)Cross-Validation - 交叉验证•Naive Bayes - 朴素贝叶斯1)Naive Bayes - 朴素贝叶斯2)Bayes' Theorem - 贝叶斯定理3)Prior Probability - 先验概率4)Posterior Probability - 后验概率5)Likelihood - 似然6)Class Conditional Probability - 类条件概率7)Feature Independence Assumption - 特征独立假设8)Multinomial Naive Bayes - 多项式朴素贝叶斯9)Gaussian Naive Bayes - 高斯朴素贝叶斯10)Bernoulli Naive Bayes - 伯努利朴素贝叶斯11)Laplace Smoothing - 拉普拉斯平滑12)Add-One Smoothing - 加一平滑13)Maximum A Posteriori (MAP) - 最大后验概率14)Maximum Likelihood Estimation (MLE) - 最大似然估计15)Classification - 分类16)Feature Vectors - 特征向量17)Training Set - 训练集18)Test Set - 测试集19)Class Label - 类标签20)Confusion Matrix - 混淆矩阵•Clustering - 聚类1)Clustering - 聚类2)Centroid - 质心3)Cluster Analysis - 聚类分析4)Partitioning Clustering - 划分式聚类5)Hierarchical Clustering - 层次聚类6)Density-Based Clustering - 基于密度的聚类7)K-Means Clustering - K均值聚类8)K-Medoids Clustering - K中心点聚类9)DBSCAN (Density-Based Spatial Clustering of Applications with Noise) - 基于密度的空间聚类算法10)Agglomerative Clustering - 聚合式聚类11)Dendrogram - 系统树图12)Silhouette Score - 轮廓系数13)Elbow Method - 肘部法则14)Clustering Validation - 聚类验证15)Intra-cluster Distance - 类内距离16)Inter-cluster Distance - 类间距离17)Cluster Cohesion - 类内连贯性18)Cluster Separation - 类间分离度19)Cluster Assignment - 聚类分配20)Cluster Label - 聚类标签•K-Means - K-均值1)K-Means - K-均值2)Centroid - 质心3)Cluster - 聚类4)Cluster Center - 聚类中心5)Cluster Assignment - 聚类分配6)Cluster Analysis - 聚类分析7)K Value - K值8)Elbow Method - 肘部法则9)Inertia - 惯性10)Silhouette Score - 轮廓系数11)Convergence - 收敛12)Initialization - 初始化13)Euclidean Distance - 欧氏距离14)Manhattan Distance - 曼哈顿距离15)Distance Metric - 距离度量16)Cluster Radius - 聚类半径17)Within-Cluster Variation - 类内变异18)Cluster Quality - 聚类质量19)Clustering Algorithm - 聚类算法20)Clustering Validation - 聚类验证•Dimensionality Reduction - 降维1)Dimensionality Reduction - 降维2)Feature Extraction - 特征提取3)Feature Selection - 特征选择4)Principal Component Analysis (PCA) - 主成分分析5)Singular Value Decomposition (SVD) - 奇异值分解6)Linear Discriminant Analysis (LDA) - 线性判别分析7)t-Distributed Stochastic Neighbor Embedding (t-SNE) - t-分布随机邻域嵌入8)Autoencoder - 自编码器9)Manifold Learning - 流形学习10)Locally Linear Embedding (LLE) - 局部线性嵌入11)Isomap - 等度量映射12)Uniform Manifold Approximation and Projection (UMAP) - 均匀流形逼近与投影13)Kernel PCA - 核主成分分析14)Non-negative Matrix Factorization (NMF) - 非负矩阵分解15)Independent Component Analysis (ICA) - 独立成分分析16)Variational Autoencoder (VAE) - 变分自编码器17)Sparse Coding - 稀疏编码18)Random Projection - 随机投影19)Neighborhood Preserving Embedding (NPE) - 保持邻域结构的嵌入20)Curvilinear Component Analysis (CCA) - 曲线成分分析•Principal Component Analysis (PCA) - 主成分分析1)Principal Component Analysis (PCA) - 主成分分析2)Eigenvector - 特征向量3)Eigenvalue - 特征值4)Covariance Matrix - 协方差矩阵。

Spatio-Temporal LSTM with Trust Gates for3D Human Action Recognition817 respectively,and utilized a SVM classifier to classify the actions.A skeleton-based dictionary learning utilizing group sparsity and geometry constraint was also proposed by[8].An angular skeletal representation over the tree-structured set of joints was introduced in[9],which calculated the similarity of these fea-tures over temporal dimension to build the global representation of the action samples and fed them to SVM forfinal classification.Recurrent neural networks(RNNs)which are a variant of neural nets for handling sequential data with variable length,have been successfully applied to language modeling[10–12],image captioning[13,14],video analysis[15–24], human re-identification[25,26],and RGB-based action recognition[27–29].They also have achieved promising performance in3D action recognition[30–32].Existing RNN-based3D action recognition methods mainly model the long-term contextual information in the temporal domain to represent motion-based dynamics.However,there is also strong dependency between joints in the spatial domain.And the spatial configuration of joints in video frames can be highly discriminative for3D action recognition task.In this paper,we propose a spatio-temporal long short-term memory(ST-LSTM)network which extends the traditional LSTM-based learning to two con-current domains(temporal and spatial domains).Each joint receives contextual information from neighboring joints and also from previous frames to encode the spatio-temporal context.Human body joints are not naturally arranged in a chain,therefore feeding a simple chain of joints to a sequence learner can-not perform well.Instead,a tree-like graph can better represent the adjacency properties between the joints in the skeletal data.Hence,we also propose a tree structure based skeleton traversal method to explore the kinematic relationship between the joints for better spatial dependency modeling.In addition,since the acquisition of depth sensors is not always accurate,we further improve the design of the ST-LSTM by adding a new gating function, so called“trust gate”,to analyze the reliability of the input data at each spatio-temporal step and give better insight to the network about when to update, forget,or remember the contents of the internal memory cell as the representa-tion of long-term context information.The contributions of this paper are:(1)spatio-temporal design of LSTM networks for3D action recognition,(2)a skeleton-based tree traversal technique to feed the structure of the skeleton data into a sequential LSTM,(3)improving the design of the ST-LSTM by adding the trust gate,and(4)achieving state-of-the-art performance on all the evaluated datasets.2Related WorkHuman action recognition using3D skeleton information is explored in different aspects during recent years[33–50].In this section,we limit our review to more recent RNN-based and LSTM-based approaches.HBRNN[30]applied bidirectional RNNs in a novel hierarchical fashion.They divided the entire skeleton tofive major groups of joints and each group was fedSpatio-Temporal LSTM with Trust Gates for3D Human Action RecognitionJun Liu1,Amir Shahroudy1,Dong Xu2,and Gang Wang1(B)1School of Electrical and Electronic Engineering,Nanyang Technological University,Singapore,Singapore{jliu029,amir3,wanggang}@.sg2School of Electrical and Information Engineering,University of Sydney,Sydney,Australia******************.auAbstract.3D action recognition–analysis of human actions based on3D skeleton data–becomes popular recently due to its succinctness,robustness,and view-invariant representation.Recent attempts on thisproblem suggested to develop RNN-based learning methods to model thecontextual dependency in the temporal domain.In this paper,we extendthis idea to spatio-temporal domains to analyze the hidden sources ofaction-related information within the input data over both domains con-currently.Inspired by the graphical structure of the human skeleton,wefurther propose a more powerful tree-structure based traversal method.To handle the noise and occlusion in3D skeleton data,we introduce newgating mechanism within LSTM to learn the reliability of the sequentialinput data and accordingly adjust its effect on updating the long-termcontext information stored in the memory cell.Our method achievesstate-of-the-art performance on4challenging benchmark datasets for3D human action analysis.Keywords:3D action recognition·Recurrent neural networks·Longshort-term memory·Trust gate·Spatio-temporal analysis1IntroductionIn recent years,action recognition based on the locations of major joints of the body in3D space has attracted a lot of attention.Different feature extraction and classifier learning approaches are studied for3D action recognition[1–3].For example,Yang and Tian[4]represented the static postures and the dynamics of the motion patterns via eigenjoints and utilized a Na¨ıve-Bayes-Nearest-Neighbor classifier learning.A HMM was applied by[5]for modeling the temporal dynam-ics of the actions over a histogram-based representation of3D joint locations. Evangelidis et al.[6]learned a GMM over the Fisher kernel representation of a succinct skeletal feature,called skeletal quads.Vemulapalli et al.[7]represented the skeleton configurations and actions as points and curves in a Lie group c Springer International Publishing AG2016B.Leibe et al.(Eds.):ECCV2016,Part III,LNCS9907,pp.816–833,2016.DOI:10.1007/978-3-319-46487-950。

Research Proposal for MS (by Research) and/or PhD aKnowledge Representation & Memory Functionality in Human Brain using ALNNs & fMRI1Biswa SenguptaUniversity of York, EnglandMind. A mysterious form of matter secreted by the brain. Its chief activity consists in the endeavour to ascertain its own nature; the futility of the attempt being due to the fact that it has nothing but itself toknow itself with.-- Ambrose Bierce A bstractIt can be argued that generally accepted methodologies of Artificial Intelligence research are limited in the proportion of human level intelligence they can be expected to emulate. This proposal aims to understand knowledge/data representation in human brain along with understanding memory modules through the usage of Artificial Life Neural Networks (ALNNs) & fMRI data from brain scans and eventually simulating the algorithm produced, if possible, on a Cellular Automata Machine (CAM) or a similar structure. This research would add a new paradigm to evolvable neural network research & machine learning techniques presently available. A principle tenet of my methodology is to build & test real robotic systems based on the work envisaged.IntroductionDoctors see man as a neurological and biological system. Mathematicians consider man a collection of logic and computational devices. Whereas, computer experts call them interactive robots. Most of today s application is just superficial application of logic developed by the human s way of doing things. What we now require to meet the challenge of these unpredictable and confusing times is a new paradigm to guide a new age. The implicit dream of AI has been to build human level intelligence. Though building a humanoid robot is challenging but recent progress in many fields shows that it is practical to make serious attempts at this goal.The research aims to take a sub-symbolic (by saying this I am not ruling out the features of symbolic AI that tends to be helpful at times) knowledge representation (cybernetic intelligence) for problem solving techniques in designing intelligent machines and control of complex systems. It will mainly encompass non-linear and optimal control of distributed & intelligent systems through designing novel neural networks. Information processing and memory elements in human brain will be the prima facie of this research. The idea moves from the contention that matter is merely a manifestation of energy. The problem is that we have no evidence for a non-material thinking substance that survives death. The problem of consciousness has been called the mind-brain problem or the ontological problem. This idea generally encompasses the popular dualism philosophy of the mind. Paul Churchland at the University of California (San Diego) calls this interactionist property dualism. Though we shouldn t entirely rule out materialism, which suggest that the brain enables the mind. What I mean that qualia, which many cognitive scientists explain as souls of experiences is a matter. Qualia is specialised perceptual & cognitive capacity we humans enjoy. Let me cite an example, we would not be overwhelmed if we happen to remember an incident of the past which may have occurred many years ago. Hence huge amount of text, sounds & video is stored in our 150cc brain. But we need to take into account that humans do not construct a full monolithic model of the environment. There is a need to better understand the data mining information model and the phenomenon of memory in the brain.1 I am yet to start the final year of my undergraduate class hence, the views expressed here fall short of the understanding that I would develop during my dissertation on Effect of Synaptic Plasticity on CA1/CA3 hippocampal pyramidal neurons. Also this research may be a bit too optimistic for a single PhD project, so we will use the historic divide and conquer rule for this.Problem Statement & Research QuestionsThe question is Can we emulate this on the neural network structures? Perhaps not. This is where Artificial Life Neural Networks (ALNNs) become useful. They live in physical networks or in other words they are ecological networks. In ALNNs, the physical environment assigns semantics to the output of the nervous system. The behaviour of ALNNs is the result of the output of the network itself, the environment, and of their interactions. I am deeply engrossed with the thought of using ALNN in an effort to understand data representation and memory in human brain. It would be fascinating to know that the input of an ALNN primarily encodes the state of the local physical environment around the organism. There the physical environment would be of par importance in assigning a semantic to the activation pattern of the input and the output of the neural network. Eventually, ALNNs may acquire through evolution an ability to extract from the environment, reinforcement learning signals or auto-generated teaching inputs and use them to adapt to the environment during their lifetime. Most ANN models are far from the biological reality; modelling neurons and synapses and developing computational tolls able to perform efficiently in perceptive tasks are different businesses despite a common inspiration.I envisage implementing the outcome and understanding of this research, if possible in the form of an algorithm on a cellular automata machine (CAM), which presently would be a RAM based lookup table hardware device [Toffoli & Margolus 1990], using Evolutionary & Genetic Algorithms.This research would add an insight to other research that aims to understand the human brain in greater depth thus implementing the mind electronically. It will add to our existing knowledge in making clever robots & add to the present family of robots like MIT s COG & KISMET (though Minsky explains that the researchers are not very candid about the limitations of the performance of their systems) and Sony s Aibo. Such an artificial nervous system will be too complex to be humanly designable, but it may be possible to evolve it, and incrementally, by adding neural modules to an already functional artificial nervous system.Theoretical and experimental researches have a reciprocal relationship theories suggest experiments, while experiments confirm or disconfirm theories and suggest new bases for theories. Second, the data that has already been collected clearly demonstrates to an impartial observer that the phenomenon exists, so as far as the idea is concerned looking for further proof of existence is sterile. As the rising flood reaches more populated heights, machines will begin to do well in areas a greater number can appreciate. The visceral sense of a thinking presence in machinery will become increasingly widespread. The presence of minds in machines will then become self-evident.I plan to decipher knowledge representation in brain using the already available information on layering, hyper-columnarisation, neurochemical modulation, splitting neurons, retina/LGN, thalamus & V1, early associate visual cortex, temporal lobe/hypothalamus, hippocampus (for memory representation), etc. If time permits I foresee a need of constructing multi-module expansion for ALNN with automatic training. Over time, artificial nervous systems should grow in complexity, until they can be called emotional machines. An alternative approach of using a CAM is to use Eldridge and Hutchings run-time reconfigurable (FPGA) hardware system (RRANN) to execute a back-propagation learning algorithm in a feed-forward neural network.Review of Related ResearchThis research is certain to revolutionize the field of neural networks and artificial life, because it will provide a powerful new tool to evolve artificial brains with billions of neurons, and at electronic speeds. This will help to produce the Darwin Machine, i.e. a machine that evolves its own architecture. To the best of my knowledge, I wouldn t deny that similar research is going in different labs across the world but MIT s Media Lab & the AI Lab (especially the Living Machines group) require a special mention. The second is in Switzerland at the EPFL under Eduardo Sanchez & Prof. Dario Floreano. Belgium s Lernout & Hauspie (L&H) is using a CAM for speech processing feasibility studies. Several scientists acrossBelgium, Japan, Poland, China & the US are known to have been looking on related aspects of this research. I think at some point in time, I have to visit the Delayed Pointer Neural Net (DePo NN) based on Collect & Distribute Neural Net (CoDi) model & the electronic Learning Evolutionary Model (eLEM)along with Prof. Tom Mitchellimages. Terry Sejnowski at Salk Institute, has done a lot of work from modelling the hippocampus, to face recognition to speech recognition to motion perception, the latest being independent component analysis & temporal hebbian learning.Deliverable and Program Schedule (Course of Action)I have had this ambition to understand the brain since I was a grade 8 student. My undergraduate degree in Electronics & Computer Engineering would have an impetus to my attained knowledge on belief networks in dynamic systems. I am confident that my understanding of cognition modules through a signalling point of view in my undergraduate dissertation will help me to develop the relevant models for data (as an abstract of feelings) & memory in human brain. I plan to complete the first phase of my research i.e., to develop an algorithm to mimic this representation by the end of my MS & then I would like to continue implementing the outcome of my research in due course of my PhD. This will allow researchers involved with decision-making & machine learning to better understand the state space on which the former is implemented, constrained by completely noisy, constantly changing environment. Research Design, Methodology & ApproachFigure 1 Proposed model for the researchI plan to use fMRI (functional Magnetic Resonance Imaging) data with a machine learning perspective for my research. An fMRI plot produces time-series data that represent brain activity in a collection of 2D slices of the brain. Multiple 2D slices can be captured, forming a 3D image that may contain on the order of 15,000 voxels. The resulting fMRI time series thus provides us with a high resolution 3D movie of the activation across a large fraction of the brain. This would be done by automatically classifying the instantaneous cognitive state of a human being, given his/her observed fMRI activity at a single time instant or time interval. But I see some problems like cognitive activity can change within 20 milliseconds, while a single 3D scan commonly takes 3 seconds, by which point dozens if not hundreds of different cognitive events could have occurred. This problem is compounded by the activation in a timepoint having been caused by a neurophysiologic event around 8 seconds earlier (since blood oxygen levels what fMRI actually records take this long to build up). Second of all, given that the human brain is parallel, and given that we are incredibly complex in our thinking patterns, the exact same task could be carried out ( in terms of cognitive states at least) quite differently on different occasions within subjects and between subjects, even if performance is identical. Keeping these perils in mind, I would use classifiers like Gaussian naïve Bayes (GNB) as a tool for decoding and tracking the sequence of hidden cognitive states of a subject. A large success of this research relies on machine learning algorithms that can successfully learn the spatial-temporal fMRI patterns that distinguish these cognitive states.For this we should clarify our understanding of how the brain and nerves actually generate thought and language. We understand the brain at the low level, the individual neurons, and we also have the capability to understand the high level, in the sense of which part lights up when I am writing this proposal. I am aware of the limitations that can arise when trying to emulate complex thought processes but am optimistic on my capabilities.Though I have mentioned my view, to implement the algorithm on a CAM, I am pretty unsure if this becomes a non-result oriented method in the future. I plan to switch to other methodology if this approach looks pessimistic. I have concluded that this research is interesting and ambitious, but in so far as it is a life-long project (maybe several life-times) I think I need to make it even more ambitious, for the simple reason that the system I am trying to understand has many levels of abstraction and I don t think any of them can be fully understood without the others. BUT one doesn t have to understand all levels equally well, and for some questions one can ignore the lowest levels. Likewise, biological evolution evolved designs at different levels of abstraction. They co-evolved. Some aspects of the higher-level machines are implementation independent and some implementation dependent. So, it would be a good idea to understand the highest level of cognitive function first.I agree with what Piaget discovered that for example the architecture of the mind of a 3 year old child is very different from that of a typical adult, and may be even adult architectures can vary, depending on the culture, personal development trajectories, trauma, etc.The proposed research can be outlined as in Figure 1. The basic questions that I wish to explore are: How the human brain stores experiences of past events by diving them into fragments?How neuro-transmitters help the process (using perhaps action-potential)?When presented with relevant information, how does the brain co-relate the input i.e., the essential link between vision & audio to match that information?What sort of an algorithm that can mimic this behaviour? Hence I am basically heading towards brain-based knowledge representation architecture.I am still building up other perspectives not limiting only to levels that are relatively close to brain structure. For this I am learning and reading on Central Nervous System as neuroscience has been developing rapidly over the last 25 years. I am making an effort to understand the organization of the real nervous system, and, along with that, the organization of behaviours as seen in the field of ethology, where there are interesting studies of behavioural evolution which is fundamental to understanding brain evolution. Also at present, my time is adsorbed on maximum likelihood, information theory & expectation maximization (esp. Baum s algorithm) apart from enlightening myself with probability theory, pattern recognition and signal processing.Constructing a mind is simply a different kind of problem of how we can synthesise organizational systems that can support a large enough diversity of different schemes, yet enable them to work together to exploit one another s abilities. I gaze to the question of self-organization, whether some processes can dissipate energy (computer cycles) and locally reverse entropy (get more & more complex). In other words, what kind of system can eventually produce a brain, rather than what is the brain.I will not rule out the fact that I might come across techniques that may change my angle of attack towards the problem and help in formation of a robust algorithm. I presume that my multitude encounters with hardcore research in the British Aerospace DCSC lab in the form of developing scenario based assessment (for software reliability) to produce an algorithm using neural network & genetic algorithms, will help my capability as a prospective researcher. The work here developed my critical understanding of search space. Especially, my knowledge on meta-heuristic search along with reliability perspective of software using Markov Analysis, Queuing Networks, Stochastic Petri Nets (SPN), etc to name a few have been greatly enlightened here. I have also thought of taking an initiative to develop parallel search spaces, but it needs a bit more thought.I am indebted to Marvin Minsky (co-founder MIT AI Labs), Aaron Sloman (University of Birmingham), Gerald E. Schneider (MIT Brain & Cognitive Science) & Jordan Pollack (Brandeis) for increasing my productivity by their useful inputs during the writing of this research proposal.References1.Parisi D., Cecconi F., and Nolfi S. (1990), Econets: Neural networks that learn in an environment.Network, 1: 149-1682.Harnard S. (1990), Symbol grounding problem. Physica D, 42: 335-4643.Ackley D.E., and Littman M.L. (1991), Interaction between learning and evolution. In C.G. Langton et.al(Eds) Proceedings of the second conference on Artificial Life. Addison-Wesley: Reading, MA4.Nolfi S, Parisi D. (1993), Auto-teaching networks that develop their own teaching input, In J.L.Deneubourg, H. Bersini, S. Goss, G. Nicholis, R. Dagonnier (Eds), Proceedings of the second European Conference on Artificial Life, Brussels, Free University of Brussels5.Nolfi S, Parisi D., Neural Networks in an Artificial Life Perspective6.Brooks R.A., Prospects for Human Level Intelligence for Humanoid Robots7.Garis Hugo de, CAM-Brain The evolutionary Engineering of a Billion Neuron Artificial Brain by 2001which Grows/Evolves at Electronic speeds inside a cellular Automata Machine (CAM)8.Gazzaniga Michael S. et al., Cognitive Neuroscience The Biology of the mind9.Nichols John G. et al., From Neuron to Brain10.Mc Clelland James L, Understanding the Mind by simulating the Brain11.Korkin M and de Garis H, The CAM Brian Machine (CBM) An FPGA based hardware Tool that evolvesa 1000 neuron-net circuit module in seconds and updates a 75 million neuron artificial brain for real-timerobot control12.Dinerstein J, Dinerstein N and de Garis H, Automatic Multi-module neural network evolution in anartificial brain13.Brooks R.A, Breazeal C, Robert I, Kemp C.C, Marjanovic M, Scassellati B, Williamson M.M, AlternativeEssences of Intelligence14.Yao X, Evolving Artificial Neural Networks15.Mitchell Tom M. et al, Classifying Instantaneous Cognitive States from fMRI Data16.Minsky M. (1987), The Society of Mind, Simon and Schuster17.Minsky M., Logical vs. Analogical OR Symbolic vs. Connectionist OR Neat vs. Scruffy18.Verleysen M., The explanatory power of Artificial Neural Networksa For the latest version of the proposal please refer to http:// /~bs125 [Revision 2.1]。

⼀⽂打尽知识图谱（超级⼲货，建议收藏！）©原创作者 | 朱林01 序⾔知识是⼈类在实践中认识客观世界的结晶。

知识图谱（Knowledge Graph, KG）是知识⼯程的重要分⽀之⼀，它以符号形式结构化地描述了物理世界中的概念及其相互关系。

知识图谱的基本组成形式为<实体,关系,实体>的三元组，实体间通过关系相互联结，构成了复杂的⽹状知识结构。

图1 知识图谱组成复杂的⽹状知识结构知识图谱从萌芽思想的提出到如今已经发展了六⼗多年，衍⽣出了许多独⽴的研究⽅向，并在众多实际⼯程项⽬和⼤型系统中发挥着不可替代的重要作⽤。

如今，知识图谱已经成为认知和⼈⼯智能⽇益流⾏的研究⽅向，受到学术界和⼯业界的⾼度重视。

本⽂对知识图谱的历史、定义、研究⽅向、未来发展、数据集和开源库进⾏了全⾯的梳理总结，值得收藏。

02 简史图2 知识库简史图2展⽰了知识图谱及其相关概念和系统的历史沿⾰，其在逻辑和⼈⼯智能领域经历了漫长的发展历程。

图形化知识表征（Knowledge Representation）的思想最早可以追溯到1956年，由Richens⾸先提出了语义⽹（Semantic Net）的概念。

逻辑符号的知识表⽰形式可以追溯到1959年的通⽤问题求解器（General Problem Solver, GPS）。

20世纪70年代，专家系统⼀度成为研究热点，基于知识推理和问题求解器的MYCIN系统是当时最著名的基于规则的医学诊断专家系统之⼀，该专家系统知识库拥有约600条医学规则。

此后，20世纪80年代早期，知识表征经历了Frame-based Languages、KL-ONE Frame Language的混合发展时期。

⼤约在这个时期结束时的1984年，Cyc项⽬出现了，该项⽬最开始的⽬标是将上百万条知识编码成机器可⽤的形式，⽤以表⽰⼈类常识，为此专门设计了专⽤的知识表⽰语⾔CycL，这种知识表⽰语⾔是基于⼀阶关系的。

Temporal Adverbials, Negation, and the Bangla PerfectWe establish that the perfect in Bangla has an unusual restriction: it does not allow adverbs to modify the reference time. We propose a syntactic account and we further suggest that another puzzling fact about the perfect in Bangla – that it cannot be negated (Ramchand 2005) – stems from the prohibition against reference time modification.Adverbial Modification. The past perfect in several languages is ambiguous when modified by so-called ‘positional’ temporal adverbials, i.e., adverbials that make reference to specific time intervals (e.g., McCoard 1978, Giorgi and Pianesi 1998, Musan 2001). In (1) the adverbial can restrict either the time interval at which the event holds – the event time (ET), or the time interval from the perspective of which the event is described – the reference time (RT), (ignoring the issue of how the two readings correlate with word order). Similar ambiguities obtain with the present perfect, see (2). In contrast, the Bangla perfect does not allow RT modification: (3) and (4) only have an ET modification reading – the submission happened on Sunday/today.(1) (On Sunday) Rick had submitted the homework (on Sunday). √ET√RT(2) (Today) Rick has submitted the homework (today). √ET√RT(3) robibare rik homwark jOma kor-e-ch-il-o √ET * RTSunday-loc Rick homework submission do-e-ch-past-3‘Rick had submitted the homework on Sunday.’(4) aj(-ke) rik homwark jOma kor-e-ch-e √ET* RTtoday Rick homework submission do-e-ch-3‘Rick has submitted the homework today.’The -e-ch forms are perfects.Could the -e-ch forms in Bangla, as in (3) and (4), be simple tenses rather than perfects, thus accounting for the absence of ambiguity of adverbial modification? Several facts reveal that this is not so: (i) the present perfect allows modification by now, while the past progressive and the simple past do not, suggesting that the present perfect is not simply another past tense form (see (5)); (ii)in embedded clauses, the present perfect requires the ET to precede a past RT introduced by the matrix tense, as in (6) and (7), suggesting that it does not behave as a present tense (it could still, of course, be like a simple past, in a language without sequence of tense); (iii) person marking varies with tense; the present perfect inflects as a present tense and the past perfect inflects as a past tense (cf. the 3 person kor-e-ch-e ‘has done’, kor-ch-e‘is doing’, kOr-e‘does’; vs. kor-e-ch-il-o ‘had done’, kor-ch-il-o ‘was doing’, kor-l-o‘did’). Finally, the -e-ch forms are considered perfects in Chatterji (1926), Chattopadhyay (1988), and Ramchand (2004). Thus, the puzzle of adverbial modification is real.(5) ekhon rik homwark jOma { kor-e-ch-e / * kor-ch-il-o / * kor-l-o }now Rick homework submission do-e-ch-3 do-ch-past-3 do-past-3‘Rick {has submitted / * was submitting / *submitted} the homework now’(6) ami baRi eS-e jan-l-am je Se eS-e-ch-il-oI home come-e know-past-1 that he come-e-ch-past-3‘Having come home, I knew that he had come.’ (Chattopadhyay 1988: 22)(7) ami bol-l-am o LA-te thek-e-ch-eI say-past-1 he LA-loc stay-e-ch-3‘I said he lived in LA.’ (only precedence, no simultaneous reading)1Analysis. The affix -ch, a remnant of the auxiliary verb ach- ‘be’ (Lahiri 2000, Butt and Lahiri 2002) spells out a semantically vacuous functional item that embeds PERFECT(and also IMPERFECTIVE, as in kor-ch-il-o ‘was doing’, but we put this aside). See (8) for a hierarchical representation (ignoring word order).(8) [T ENSE[-ch[PERFECT[VIEWPOINT ASPECT[v P ]]]]]The lexical semantics of PERFECT is as in (9), which follows Pancheva and von Stechow (2004) in treating the PERFECT as a weak relative past: it introduces an interval no part of which may follow the reference time introduced by TENSE.(9) [[PERFECT]] = λp<i,t> λt i ∃t′i [t′≤ t & p(t′)] (t′≤ t iff there is no t″⊂ t′, s.t. t″ > t)The affix -e, both on its own, e.g., baRi eS-e ‘having come’ in (6), and in combination with -ch in the perfect, marks RESULTATIVE viewpoint; see (10) for its semantics. The composition of PERFECT and RESULTATIVE yields the needed semantics for Bangla perfects, which lack universal readings (see also Ramchand 2005).(10) [[RESULTATIVE]]= λP<v,t> λt i∃s∃e [t ⊂τ(s) & s is a target state of e & P(e)]The PERFECT moves to the affix –ch and then to T ENSE; this syntax precludes adverbs from being merged and interpreted higher than PERFECT. Accordingly, the LF in (11a) is not possible; only the one in (11b) is. (11a) derives RT modification (see (12a), and it is not available in the Bangla perfect. (11b) is the LF behind ET modification (see (12b), and it is the only structure available in the Bangla perfect. Thus, we account for the restriction on temporal modification in (3)-(4).(11) a.*[T ENSE - ch [adverbial[PERFECT[RESULTATIVE →-e [v P ]]]]]]b. [T ENSE-ch -PERFECT[adverbial[RESULTATIVE →-e [v P ]]]]]](12) a. * ∃t [t < t c & t ⊆Sunday & ∃t′ [t′≤ t & ∃s∃e [t′⊂τ(s) & s is a target state of e & P(e)]]]b. ∃t [t < t c & ∃t′ [t′≤ t & t′⊆Sunday & ∃s∃e [t′⊂τ(s) & s is a target state of e & P(e)]]Negation and the perfect. We further suggest that the prohibition against RT modification in the perfect is responsible for the fact that the perfect cannot be negated. The negative marker na combines freely with the simple past and present, the past and present progressive, and the past habitual – all tense-aspect forms except for the perfects (Ramchand 2005), see (13) for some representative examples from the non-perfect tense forms. However, the perfect cannot appear with na. Instead of the ungrammatical (14a) we get (14b), where the verb is not explicitly marked for tense and aspect, but is interpreted as past.(13) ami am-Ta { khe-l-am / kha-cch-i / kha-cch-il-am } (na)I mango-cl eat-pst-1 eat-ch-1 eat-ch-pst-1 NEG‘I {did (not) eat / am (not) eating / was (not) eating} the mango.’(14) a. * ami am-Ta { khe-ye-ch-i / khe-ye-ch-il-am} naI mango-class eat-e-ch-1 eat-e-ch-pst-1 NEG‘I {have / had} not eaten the mango.’b. ami am-Ta kha-i-niI mango-class eat -1 -NEG‘I didn’t eat the mango.’The proposal that the na negation in Bangla is a reference time modifier is consistent with the semantics proposed by Ramchand (2005). It is a negative existential quantifier over events asserting that no event of the relevant kind occurs within a specified time interval,i.e., the RT.2。