An Experimental Study of the Potential of Using Small

Slides:



Advertisements
Similar presentations
Learning Outcomes Participants will be able to analyze assessments
Advertisements

Musical Background - Pianos are extremely complicated (over 9,000 pieces) - Highly certified piano technicians can take days to prepare a concert piano.
Attention Deficit Hyperactivity Disorder (ADHD) Student Classification Using Genetic Algorithm and Artificial Neural Network S. Yenaeng 1, S. Saelee 2.
More on Data Mining KDnuggets Datanami ACM SIGKDD
AICT5 – eProject Project Planning for ICT. Process Centre receives Scenario Group Work Scenario on website in October Assessment Window Individual Work.
A Genetic Algorithms Approach to Feature Subset Selection Problem by Hasan Doğu TAŞKIRAN CS 550 – Machine Learning Workshop Department of Computer Engineering.
1 Commissioned by PAMSA and German Technical Co-Operation National Certificate in Paper & Pulp Manufacturing NQF Level 3 Collect and use data to establish.
COMPARISON OF IMAGE ANALYSIS FOR THAI HANDWRITTEN CHARACTER RECOGNITION Olarik Surinta, chatklaw Jareanpon Department of Management Information System.
Effective Project Management: Traditional, Agile, Extreme Presented by (facilitator name) Managing Complexity in the Face of Uncertainty Ch08: How to Close.
Disclosure risk when responding to queries with deterministic guarantees Krish Muralidhar University of Kentucky Rathindra Sarathy Oklahoma State University.
A hybrid SOFM-SVR with a filter-based feature selection for stock market forecasting Huang, C. L. & Tsai, C. Y. Expert Systems with Applications 2008.
Indirect Supervision Protocols for Learning in Natural Language Processing II. Learning by Inventing Binary Labels This work is supported by DARPA funding.
Human pose recognition from depth image MS Research Cambridge.
CS532 TERM PAPER MEASUREMENT IN SOFTWARE ENGINEERING NAVEEN KUMAR SOMA.
Tell Me What You See and I will Show You Where It Is Jia Xu 1 Alexander G. Schwing 2 Raquel Urtasun 2,3 1 University of Wisconsin-Madison 2 University.
A NOVEL METHOD FOR COLOR FACE RECOGNITION USING KNN CLASSIFIER
Improved Video Categorization from Text Metadata and User Comments ACM SIGIR 2011:Research and development in Information Retrieval - Katja Filippova -
Predicting the Location and Time of Mobile Phone Users by Using Sequential Pattern Mining Techniques Mert Özer, Ilkcan Keles, Ismail Hakki Toroslu, Pinar.
Learning Photographic Global Tonal Adjustment with a Database of Input / Output Image Pairs.
Workshop on Census Cartography and Management, Port-of-Spain, Trinidad and Tobago, October 2007 How to Structure, Design and Evaluate Capacity for.
Workshop on Census Cartography and Management, Bangkok, Thailand, 15–19 October 2007 How to Structure, Design and Evaluate Capacity for the use of GIS.
Workshop on International Standards, Contemporary Technologies and Regional Cooperation, Noumea, New Caledonia, 04–08 February 2008 How to Structure, Design.
ItemBased Collaborative Filtering Recommendation Algorithms 1.
Face recognition using Histograms of Oriented Gradients
Sparse Coding: A Deep Learning using Unlabeled Data for High - Level Representation Dr.G.M.Nasira R. Vidya R. P. Jaia Priyankka.
Experience Report: System Log Analysis for Anomaly Detection
Olivier Siohan David Rybach
Big data classification using neural network
Decision Support and Business Intelligence Systems
Compact Bilinear Pooling
Week 3 (June 6 – June10 , 2016) Summary :
Intro to Machine Learning
TÆMS-based Execution Architectures
Edinburgh Napier University
Hypothesis Testing, Validity, and Threats to Validity
Regularizing Face Verification Nets To Discrete-Valued Pain Regression
IS Project Management How to Close a TPM Project
Supporting Fault-Tolerance in Streaming Grid Applications
Efficient Deep Model for Monocular Road Segmentation
*current controlled assessment plans are unknown
Master’s Thesis defense Ming Du Advisor: Dr. Yi Shang
Bird-species Recognition Using Convolutional Neural Network
Introduction to Systems Analysis and Design
A Unifying View on Instance Selection
Summary Project goals related to the Census Selection of a Methodology
Face Components detection
Similarity based on Shape and Appearance
PROJECTS SUMMARY PRESNETED BY HARISH KUMAR JANUARY 10,2018.
CAR EVALUATION SIYANG CHEN ECE 539 | Dec
EPQ Learner Outcomes identify, design, plan and complete an individual project, applying a range of organisational skills and strategies to meet.
Age and Gender Classification using Convolutional Neural Networks
Hyperspectral Image Classification Using ResNeXt with Squeeze and Excitation Block Jiayu Wang.
Semantic Similarity Detection
Good View Hunting: Learning Photo Composition from Dense View Pairs Zijun Wei1, Jianming Zhang2, Xiaohui Shen2, Zhe Lin2, Radomír Měch2, Minh Hoai1, Dimitris.
STEM PROJECT Science Inquiry
Experiment No. (1) - an introduction to MATLAB
Effective Project Management: Traditional, Agile, Extreme
Soft System Stakeholder Analysis
AICT5 – eProject Project Planning for ICT
Deep Learning for the Soft Cutoff Problem
M. Kezunovic (P.I.) S. S. Luo D. Ristanovic Texas A&M University
Modeling and Analysis Tutorial
Sketch Object Prediction
The site to download BALBES:
Model Compression Joseph E. Gonzalez
Automatic Handwriting Generation
Bug Localization with Combination of Deep Learning and Information Retrieval A. N. Lam et al. International Conference on Program Comprehension 2017.
Soft System Stakeholder Analysis
Device Failure Prediction
Presentation transcript:

An Experimental Study of the Potential of Using Small Size DNNs Collaboratively Solve a Complex Task Boyu Zhang Dec. 12, 2018

Executive Summary Objective: Tasks Performed: Self Assessment: This project performs an experimental study to evaluate the potential of using small size DNN models collaboratively to solve a complex task. In particular, we studied how the search space of face ID can be narrowed by using predictions generated by age and gender models. Tasks Performed: Find the appropriate dataset and clean it up. Trained DNN models for age and gender prediction. Then evaluate their individual performance as well as how much the search space is narrowed. Aggressively simplified the trained models and evaluated them again. Self Assessment: Basically finished the tasks in the proposal but this is just a preliminary work towards the understanding of collaboration between small size DNNs.

Data Preparation Requirement of the dataset: Adience: Clean up: For each sample in the dataset, there are multiple labels associated with it, and there exist a relationship between these labels. Thus, we can train some small DNN models to predict some labels and use the predictions to infer the information of the remaining labels. Adience: This is a dataset of face photos that was intended to facilitate the study of age and gender recognition. The images included in this dataset are captured under real-world imaging conditions. 26,580 images of 2,284 persons. Each image has three labels: age (1 of 8 groups), gender (1 of 2 groups), and person ID. Clean up: The original dataset and the meta files associated with it are very dirty. They need to be cleaned before performing any training. The downloaded dataset only contains 19,370 images. Moreover, are many age and gender labels are invalid and / or missing. After cleaning up, 2,192 (output 2,284) persons have both valid age and gender labels, and this corresponds to 17,608 (output 19,370) images.

Model Training and Structural Simplification Train age and gender models: Model definitions are from the paper of the dataset. We converted the images to TFRecord format for high performance training. We implemented the data pipeline and training procedure in TensorFlow with Python. The training is performed with GPU (1080 Ti) on my desktop. Each takes about less then one hour. The trained models are then get evaluated with another Python script. Structural Simplification (neuron elimination): We extract the trained weights and intermediate outputs from the trained models. We used Matlab to carry out the neuron elimination technique. We evaluated the impact of the neuron elimination technique on model accuracies as well as the effectiveness of narrowing down the search space.

Results Training Loss Training Accuracy Validation Accuracy Age Model Gender Model

Results (simplify age model) Metrics: Gender accuracy, Age accuracy (exact / off-by-one), both accuracy (exact / off-by-one). List of Candidate size (exact / off-by-one). FC1 FC2 FC3

Results (simplify gender model) Metrics: Gender accuracy, Age accuracy (exact / off-by-one), both accuracy (exact / off-by-one). List of Candidate size (exact / off-by-one). FC1 FC2 FC3

Results (simplify both models) Metrics: Gender accuracy, Age accuracy (exact / off-by-one), both accuracy (exact / off-by-one). List of Candidate size (exact / off-by-one). FC1 FC2 FC3

Conclusions and Lesson Learned Using current age and gender predictions to narrow down the search space of person ID is not very effective. The accuracies of the age and gender models need to be further improved to achieve more effective narrow down. The age and gender models’ accuracy as well as the effectiveness of narrowing down the person ID search space are not very sensitive to the input neuron numbers. This may suggest there exist large redundancy in the original age and gender models. Lesson learned: Don’t trust the advertisements of dataset. There may exist many traps in the dataset.