DL Approaches to Time Series Data. Miro Enev, DL Solution Architect Jeff Weiss, Director West SAs

Similar documents
Anticipating Visual Representations from Unlabeled Data. Carl Vondrick, Hamed Pirsiavash, Antonio Torralba

UNSUPERVISED LEARNING

Deep learning / Ian Goodfellow, Yoshua Bengio and Aaron Courville. - Cambridge, MA ; London, Spis treści

Multimodal context analysis and prediction

Anomaly Detection in Manufacturing Systems Using Structured Neural Networks

Incorporating detractors into SVM classification

Deep Learning. Basics and Intuition. Constantin Gonzalez Principal Solutions Architect, Amazon Web Services

Real Estate Price Prediction with Regression and Classification CS 229 Autumn 2016 Project Final Report

Introduction To Artificial Neural Networks

ARTIFICIAL NEURAL NETWORK PART I HANIEH BORHANAZAD

Introduction to Machine Learning Midterm Exam

Linear Models for Regression CS534

CONTEMPORARY ANALYTICAL ECOSYSTEM PATRICK HALL, SAS INSTITUTE

Introduction to Natural Computation. Lecture 9. Multilayer Perceptrons and Backpropagation. Peter Lewis

Reservoir Computing and Echo State Networks

ECE521 Lecture 7/8. Logistic Regression

An Adaptive Multi-Modeling Approach to Solar Nowcasting

A Wavelet Neural Network Forecasting Model Based On ARIMA

(Feed-Forward) Neural Networks Dr. Hajira Jabeen, Prof. Jens Lehmann

Financial Risk and Returns Prediction with Modular Networked Learning

Deep Learning Architecture for Univariate Time Series Forecasting

Chart types and when to use them

The Changing Landscape of Land Administration

Anomaly Detection for the CERN Large Hadron Collider injection magnets

Journal of Chemical and Pharmaceutical Research, 2014, 6(5): Research Article

How to do backpropagation in a brain

A Variance Modeling Framework Based on Variational Autoencoders for Speech Enhancement

2011 Pearson Education, Inc

Prediction for night-time ventilation in Stanford s Y2E2 building

Linear Models for Regression CS534

Predicting Solar Flares by Converting GOES X-ray Data to Gramian Angular Fields (GAF) Images

Integrated Electricity Demand and Price Forecasting

Dimensionality Reduction and Principle Components Analysis

Essence of Machine Learning (and Deep Learning) Hoa M. Le Data Science Lab, HUST hoamle.github.io

Introduction to Machine Learning

Deep Sequence Models. Context Representation, Regularization, and Application to Language. Adji Bousso Dieng

FORECASTING: A REVIEW OF STATUS AND CHALLENGES. Eric Grimit and Kristin Larson 3TIER, Inc. Pacific Northwest Weather Workshop March 5-6, 2010

About Nnergix +2, More than 2,5 GW forecasted. Forecasting in 5 countries. 4 predictive technologies. More than power facilities

Predicting rock conditions ahead of the face

Visual meta-learning for planning and control

Data Mining. Chapter 1. What s it all about?

Short and medium term solar irradiance and power forecasting given high penetration and a tropical environment

STA 414/2104: Lecture 8

Discovery Through Situational Awareness

Introduction to Neural Networks

Unsupervised Neural Nets

CS325 Artificial Intelligence Chs. 18 & 4 Supervised Machine Learning (cont)

Machine Learning for Signal Processing Neural Networks Continue. Instructor: Bhiksha Raj Slides by Najim Dehak 1 Dec 2016

Introduction to Convolutional Neural Networks (CNNs)

CSC321 Lecture 16: ResNets and Attention

Memory-Augmented Attention Model for Scene Text Recognition

Dynamic Data Modeling, Recognition, and Synthesis. Rui Zhao Thesis Defense Advisor: Professor Qiang Ji

Hidden Markov Models Hamid R. Rabiee

APPLYING BIG DATA TOOLS TO ACQUIRE AND PROCESS DATA ON CITIES

Wind Energy Predictions of Small-Scale Turbine Output Using Exponential Smoothing and Feed- Forward Neural Network

22/04/2014. Economic Research

Deep Learning Lab Course 2017 (Deep Learning Practical)

A Hybrid Deep Learning Approach For Chaotic Time Series Prediction Based On Unsupervised Feature Learning

Weather Forecasting using Soft Computing and Statistical Techniques

Predict Time Series with Multiple Artificial Neural Networks

Speaker Representation and Verification Part II. by Vasileios Vasilakakis

DANIEL WILSON AND BEN CONKLIN. Integrating AI with Foundation Intelligence for Actionable Intelligence

Human-Oriented Robotics. Temporal Reasoning. Kai Arras Social Robotics Lab, University of Freiburg

1 Random walks and data

Article from. Predictive Analytics and Futurism. July 2016 Issue 13

RS Metrics CME Group Copper Futures Price Predictive Analysis Explained

Deep Learning Srihari. Deep Belief Nets. Sargur N. Srihari

Calculating Land Values by Using Advanced Statistical Approaches in Pendik

6.036 midterm review. Wednesday, March 18, 15

Interleaved Factorial Non-Homogeneous Hidden Markov Models for Energy Disaggregation

Unsupervised Feature Extraction by Time-Contrastive Learning and Nonlinear ICA

Sequence Modeling with Neural Networks

arxiv: v1 [cs.lg] 2 Feb 2019

Feature Design. Feature Design. Feature Design. & Deep Learning

Advancing Machine Learning and AI with Geography and GIS. Robert Kircher

ECE 521. Lecture 11 (not on midterm material) 13 February K-means clustering, Dimensionality reduction

From statistics to data science. BAE 815 (Fall 2017) Dr. Zifei Liu

Introduction to Machine Learning Midterm Exam Solutions

Time Series Data Cleaning

HYPERGRAPH BASED SEMI-SUPERVISED LEARNING ALGORITHMS APPLIED TO SPEECH RECOGNITION PROBLEM: A NOVEL APPROACH

Forecasting demand in the National Electricity Market. October 2017

MODELLING ENERGY DEMAND FORECASTING USING NEURAL NETWORKS WITH UNIVARIATE TIME SERIES

Bayesian Deep Learning

Effective Strategies for Forecasting a Product Hierarchy

Pattern Recognition and Machine Learning. Artificial Neural networks

Deep learning on 3D geometries. Hope Yao Design Informatics Lab Department of Mechanical and Aerospace Engineering

Variational Autoencoder

Neural Networks. Nicholas Ruozzi University of Texas at Dallas

CHAPTER 4 FAULT DIAGNOSIS OF BEARINGS DUE TO SHAFT RUB

Predicting New Search-Query Cluster Volume

Probability and Information Theory. Sargur N. Srihari

Anomaly Detection and Categorization Using Unsupervised Deep Learning

Handwritten Indic Character Recognition using Capsule Networks

Reasoning Under Uncertainty Over Time. CS 486/686: Introduction to Artificial Intelligence

CHAPTER 6 FAULT DIAGNOSIS OF UNBALANCED CNC MACHINE SPINDLE USING VIBRATION SIGNATURES-A CASE STUDY

A Feature Based Neural Network Model for Weather Forecasting

Hierarchical models for the rainfall forecast DATA MINING APPROACH

Chapter 7 Forecasting Demand

AN INTERNATIONAL SOLAR IRRADIANCE DATA INGEST SYSTEM FOR FORECASTING SOLAR POWER AND AGRICULTURAL CROP YIELDS

Transcription:

DL Approaches to Time Series Data Miro Enev, DL Solution Architect Jeff Weiss, Director West SAs

Agenda Define Time Series [ Examples & Brief Summary of Considerations ] Semi-supervised Anomaly Detection [ with Deep Autoencoders ] Ensemble event detection and classification [ with MLPs & CNN-MLPs ] Prediction [ with Dual Attentional RNNs (DA-RNNs) ]

Time Series [ One Definition ] A time series is a series of values recorded over equally spaced time intervals. The amount of time between observations is the sampling interval or sampling rate. The time series represent some underlying, partially-random ( stochastic ) process which generated the data. We want to use the data to make guesses ( inferences ) about the process, and want to make reliable guesses while being clear about the uncertainty involved. The complication is that each observation is dependent on all the other observations, and in fact its this dependence that we want to learn and draw inferences about.

Popularity of French Names

NVIDIA Stock https://www.google.com/finance?chdnp=0&chdd=0&chds=1& chdv=1&chvs=linear&chdeh=0&chfdeh=0&chdet=1494360000 000&chddm=98532&chls=IntervalBasedLine&q=NASDAQ:NVD A&ntsp=0&ei=N_kRWemBAoqxjAHRgZDwBA

Categories of Time Series Analysis Prediction/Forecasting Weather, Sales, Inventory, Financial Markets Anomaly Detection Predictive Maintenance Reconstruction Virtual Sensors Classification Event/Activity Recognition

Time Series Best Practices Think hard about the input(s) Time and frequency domain representations Lags and windowing (multi-time scale) Study correlational structure Use semi/unsupervised approaches [ free ] Feature Building, Data Buckets, Anomaly Detection Gather and/or generate ground truth Class imbalance & sample weighting Try DL fabrics with increasing complexity & Ensemble Use a non DL baseline [ ARIMA, SVMs, random forests ] Use forward chained cross-validation for evaluation Train attentional mechanism*

DL for Anomaly Detection

Deep Autoencoder Anomaly Detection 13 Sensors, 100Hz, NASA Dataset,.5 seconds window, 650 dimensions per sample [ 256, 196, 136, 76, 14 ] Anomaly Detection and Fault Disambiguation in Large Flight Data: A Multi-modal Deep Auto-encoder Approach, K. Reddy et al, United Technologies Research Center (PHM16)

Sample Reconstructions Reconstruction & Feature Learning Exceptionally low normalized RMS reconstruction error (0.04 0.09) Anomaly Detection and Fault Disambiguation in Large Flight Data: A Multi-modal Deep Auto-encoder Approach, K. Reddy et al, United Technologies Research Center (PHM16)

Anomaly Analysis 11-layer 14-dimensional bottleneck DAE yields 97.8% true positive detection rate with 0.0% false alarm Anomaly Detection and Fault Disambiguation in Large Flight Data: A Multi-modal Deep Auto-encoder Approach, K. Reddy et al, United Technologies Research Center (PHM16) Artifcially created anomalies [ Spall Fault ; Ballscrew Jam ]

Anomaly Analysis/Interpretation 11-layer 14-dimensional bottleneck DAE yields 97.8% true positive detection rate with 0.0% false alarm Anomaly Detection and Fault Disambiguation in Large Flight Data: A Multi-modal Deep Auto-encoder Approach, K. Reddy et al, United Technologies Research Center (PHM16) Artificially created anomalies [ Spall Fault ; Ballscrew Jam ]

DL for Time Series Classification

IoT Case Study Water Disaggregation

Frequency Domain Utilization

Automated Ground Truth Building Open Edges Close Edges

Event/Edge Detection

Edge Characterization, Example 1 Open Edge Close edge

Edge Characterization, Example 2 Close edge Open Edge

DL for Time Series Prediction

Dual Attention RNNs

Performance @ NASDAQ 100 In NASDAQ 100 Stock dataset, we collect the stock prices of 81 major corporations under NASDAQ 100, which are used as the driving time series. The index value of NASDAQ 100 is used as the target series. The frequency of the data collection is one-minute. This data covers the period from July 26, 2016 to December 26, 2016, in total 104 days. Each day contains 390 data points from the opening to closing of the market. In our experiments, we use the first 90 days as the training set and the following seven days as the validation.

Performance @ SML 2010 SML 2010 is a public dataset used for indoor temperature forecasting. This dataset is collected from a monitor system mounted in a domestic house. We use room temperature as the target series and select 16 relevant driving series which contains approximately 40 days of monitoring data. The data was sampled every minute and was smoothed with 15 minute means. In our experiment, we use the first 3200 data points as the training set, the following 400 data points as the validation set, and the last 537 data points as the test set.

Attention Mechanism at Work Train Set [ NASDAQ ] Test Set [ NASDAQ ]

Thanks!