Machine Learning on Images with Noisy Human-centric Labels
When humans describe images, they have a reporting bias, in that the report only what they consider important. Thus, in addition to considering whether something is present in an image, one should consider whether it is also relevant to the image before labeling it.
Ishan Misra joins us this week to discuss his recent paper Seeing through the Human Reporting Bias: Visual Classifiers from Noisy Human-Centric Labels which explores a novel architecture for learning to distinguish presence and relevance. This work enables web-scale datasets to be useful for training, not just well groomed hand labeled corpora.
Watch on YouTube ↗
(saves to browser)
Sign in to unlock AI tutor explanation · ⚡30
Playlist
Uploads from Data Skeptic · Data Skeptic · 32 of 60
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
▶
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Data Skeptic book giveaway contest winner selection
Data Skeptic
OpenHouse - Front end and API overview
Data Skeptic
OpenHouse Crawling with AWS Lambda
Data Skeptic
[MINI] Logistic Regression on Audio Data
Data Skeptic
Data Provenance and Reproducibility with Pachyderm
Data Skeptic
[MINI] Primer on Deep Learning
Data Skeptic
Big Data Tools and Trends
Data Skeptic
[MINI] Automated Feature Engineering
Data Skeptic
The Data Refuge Project
Data Skeptic
[MINI] The Perceptron
Data Skeptic
[MINI] Feed Forward Neural Networks
Data Skeptic
Data Science at Patreon
Data Skeptic
[MINI] Backpropagation
Data Skeptic
[MINI] GPU CPU
Data Skeptic
OpenHouse
Data Skeptic
[MINI] Generative Adversarial Networks
Data Skeptic
[MINI] AdaBoost
Data Skeptic
[MINI] The Bootstrap
Data Skeptic
[MINI] Dropout
Data Skeptic
[MINI] Gini Coefficients
Data Skeptic
[MINI] Random Forest
Data Skeptic
[MINI] Heteroskedasticity
Data Skeptic
[MINI] ANOVA
Data Skeptic
Urban Congestion
Data Skeptic
[MINI] The CAP Theorem
Data Skeptic
Unstructured Data for Finance
Data Skeptic
Detecting Terrorists with Facial Recognition?
Data Skeptic
Predictive Models on Random Data
Data Skeptic
[MINI] Entropy
Data Skeptic
[MINI] F1 Score
Data Skeptic
Causal Impact
Data Skeptic
Machine Learning on Images with Noisy Human-centric Labels
Data Skeptic
The Library Problem
Data Skeptic
Stealing Models from the Cloud
Data Skeptic
Data Science at eHarmony
Data Skeptic
Multiple Comparisons and Conversion Optimization
Data Skeptic
Election Predictions
Data Skeptic
[MINI] Calculating Feature Importance
Data Skeptic
MS Connect Conference
Data Skeptic
Music21
Data Skeptic
The Police Data and the Data Driven Justice Initiatives
Data Skeptic
Studying Competition and Gender Through Chess
Data Skeptic
[MINI] Goodhart's Law
Data Skeptic
Trusting Machine Learning Models with LIME
Data Skeptic
[MINI] Leakage
Data Skeptic
Predictive Policing
Data Skeptic
Mutli-Agent Diverse Generative Adversarial Networks
Data Skeptic
[MINI] Convolutional Neural Networks
Data Skeptic
Unsupervised Depth Perception
Data Skeptic
[MINI] Max-pooling
Data Skeptic
MS Build 2017
Data Skeptic
Activation Functions
Data Skeptic
Doctor AI
Data Skeptic
[MINI] The Vanishing Gradient
Data Skeptic
CosmosDB
Data Skeptic
Estimating Sheep Pain with Facial Recognition
Data Skeptic
[MINI] Conditional Independence
Data Skeptic
MINI: Bayesian Belief Networks
Data Skeptic
Project Common Voice
Data Skeptic
[MINI] Recurrent Neural Networks
Data Skeptic
More on: Supervised Learning
View skill →Related AI Lessons
⚡
⚡
⚡
⚡
The Idempotency Nightmare in AI Pipelines: Data Loss and Recovery
Dev.to AI
My Experience with Network Anomaly Detection Using 5 Different ML Approaches
Medium · Machine Learning
My Experience with Network Anomaly Detection Using 5 Different ML Approaches
Medium · Cybersecurity
Sujar Henry on Why Access Still Isn’t Enough in Tech
Medium · Machine Learning
🎓
Tutor Explanation
DeepCamp AI