SlideShare a Scribd company logo
Machine Learning Algorithms
(MLAs) and Novelty Detection
Rangaprasad Sampath (ranga.sampath@gmail.com)
Madhusoodhana Chari S (madhucharis@gmail.com)
Teaching a Child and a MLA (Machine
Learning Algorithm)
Orange
Apple
Responses to Unseen/Novel Objects
Insights
Humans may not classify Unseen/Novel objects into known classes.
MLAs classify Unseen/Novel objects into the most probable class.
Orange Orange (30%), Apple (70%)
Apple
I don’t know Orange (72%), Apple (28%)
No response
Decision Tree Classifier
Decision Tree Classifier labels as a Car or Not based on the
measurements of length (x2) and width (x1) of a vehicle.
4
x1
Width
(m)
x2
Length
(m)
Label
4 9 !Car
4 10 !Car
4 12 Car
3 11 Car
7 10 !Car
8 16 Car
12 18 Car
10 18 Car
Training Data
Novelty Classification
Test Data that is very different from what has
been seen during training i.e. Novelty, is
classified into one of the known classes.
5
x1 x2 Label
4 9 !Car
4 10 !Car
4 12 Car
3 11 Car
Test Data (Red dots)
Recognition of an Unknown class
Proposal: Label test data points that
represent Novelty as Unknown.
6
x1 x2 Label
4 9 Unknown
4 10 Unknown
4 12 Unknown
3 11 Unknown
Test Data (Red dots) Unknown
Unknown
Unknown
Unknown
Benefits
• Avoidance – of Novel data misclassification.
• Utility – in domains where it is difficult to obtain fully representative
datasets for training.
• Retraining – could be triggered if the number of data points classified
as Unknown exceeded the number of data points classified into
known classes.
• Alerts – could be raised if the percentage of data that got classified as
an Unknown class exceeded a set threshold (say 10 %).

More Related Content

More from madhucharis

Classifier with deep deviation detection in poe iot devices
Classifier with deep deviation detection in poe iot devices Classifier with deep deviation detection in poe iot devices
Classifier with deep deviation detection in poe iot devices
madhucharis
 
Encrypted traffic malware detection twiml
Encrypted traffic malware detection twimlEncrypted traffic malware detection twiml
Encrypted traffic malware detection twiml
madhucharis
 
Long Term Evolution
Long Term EvolutionLong Term Evolution
Long Term Evolution
madhucharis
 
Weather exploratory data analysis
Weather   exploratory data analysisWeather   exploratory data analysis
Weather exploratory data analysis
madhucharis
 
IEEE Connect 2020 Novel TLS signature extraction for Encrypted malware detection
IEEE Connect 2020 Novel TLS signature extraction for Encrypted malware detectionIEEE Connect 2020 Novel TLS signature extraction for Encrypted malware detection
IEEE Connect 2020 Novel TLS signature extraction for Encrypted malware detection
madhucharis
 
Classifier with Deep Deviation Detection in PoE-IoT devices
Classifier with Deep Deviation Detection in PoE-IoT devices Classifier with Deep Deviation Detection in PoE-IoT devices
Classifier with Deep Deviation Detection in PoE-IoT devices
madhucharis
 
Researching artificial intelligence and machine learning a lgorithms final
Researching artificial intelligence and machine learning a lgorithms  finalResearching artificial intelligence and machine learning a lgorithms  final
Researching artificial intelligence and machine learning a lgorithms final
madhucharis
 
Catalyzing Innovation
Catalyzing InnovationCatalyzing Innovation
Catalyzing Innovation
madhucharis
 
Heuristics and Data Science Supervised Machine Learning
Heuristics and Data Science Supervised Machine LearningHeuristics and Data Science Supervised Machine Learning
Heuristics and Data Science Supervised Machine Learning
madhucharis
 
Heuristics Data Science Life Cycle
Heuristics Data Science Life CycleHeuristics Data Science Life Cycle
Heuristics Data Science Life Cycle
madhucharis
 
Patents
PatentsPatents
Patents
madhucharis
 
Innovation workshop
Innovation workshopInnovation workshop
Innovation workshop
madhucharis
 
Explainable AI - Posters
Explainable AI - PostersExplainable AI - Posters
Explainable AI - Posters
madhucharis
 
Learning from Data - Various Approaches - Poster
Learning from Data - Various Approaches - PosterLearning from Data - Various Approaches - Poster
Learning from Data - Various Approaches - Poster
madhucharis
 
Machine Learning and Iot change detection and security
Machine Learning and Iot change detection and securityMachine Learning and Iot change detection and security
Machine Learning and Iot change detection and security
madhucharis
 
Heavy Hitters Change Detection
Heavy Hitters Change DetectionHeavy Hitters Change Detection
Heavy Hitters Change Detection
madhucharis
 
AI Journey Poster
AI Journey PosterAI Journey Poster
AI Journey Poster
madhucharis
 
Applied NLP - Poster
Applied NLP - PosterApplied NLP - Poster
Applied NLP - Poster
madhucharis
 
Applied CNN Poster
Applied CNN PosterApplied CNN Poster
Applied CNN Poster
madhucharis
 
AI Challenges Expert Need Poster
AI Challenges Expert Need PosterAI Challenges Expert Need Poster
AI Challenges Expert Need Poster
madhucharis
 

More from madhucharis (20)

Classifier with deep deviation detection in poe iot devices
Classifier with deep deviation detection in poe iot devices Classifier with deep deviation detection in poe iot devices
Classifier with deep deviation detection in poe iot devices
 
Encrypted traffic malware detection twiml
Encrypted traffic malware detection twimlEncrypted traffic malware detection twiml
Encrypted traffic malware detection twiml
 
Long Term Evolution
Long Term EvolutionLong Term Evolution
Long Term Evolution
 
Weather exploratory data analysis
Weather   exploratory data analysisWeather   exploratory data analysis
Weather exploratory data analysis
 
IEEE Connect 2020 Novel TLS signature extraction for Encrypted malware detection
IEEE Connect 2020 Novel TLS signature extraction for Encrypted malware detectionIEEE Connect 2020 Novel TLS signature extraction for Encrypted malware detection
IEEE Connect 2020 Novel TLS signature extraction for Encrypted malware detection
 
Classifier with Deep Deviation Detection in PoE-IoT devices
Classifier with Deep Deviation Detection in PoE-IoT devices Classifier with Deep Deviation Detection in PoE-IoT devices
Classifier with Deep Deviation Detection in PoE-IoT devices
 
Researching artificial intelligence and machine learning a lgorithms final
Researching artificial intelligence and machine learning a lgorithms  finalResearching artificial intelligence and machine learning a lgorithms  final
Researching artificial intelligence and machine learning a lgorithms final
 
Catalyzing Innovation
Catalyzing InnovationCatalyzing Innovation
Catalyzing Innovation
 
Heuristics and Data Science Supervised Machine Learning
Heuristics and Data Science Supervised Machine LearningHeuristics and Data Science Supervised Machine Learning
Heuristics and Data Science Supervised Machine Learning
 
Heuristics Data Science Life Cycle
Heuristics Data Science Life CycleHeuristics Data Science Life Cycle
Heuristics Data Science Life Cycle
 
Patents
PatentsPatents
Patents
 
Innovation workshop
Innovation workshopInnovation workshop
Innovation workshop
 
Explainable AI - Posters
Explainable AI - PostersExplainable AI - Posters
Explainable AI - Posters
 
Learning from Data - Various Approaches - Poster
Learning from Data - Various Approaches - PosterLearning from Data - Various Approaches - Poster
Learning from Data - Various Approaches - Poster
 
Machine Learning and Iot change detection and security
Machine Learning and Iot change detection and securityMachine Learning and Iot change detection and security
Machine Learning and Iot change detection and security
 
Heavy Hitters Change Detection
Heavy Hitters Change DetectionHeavy Hitters Change Detection
Heavy Hitters Change Detection
 
AI Journey Poster
AI Journey PosterAI Journey Poster
AI Journey Poster
 
Applied NLP - Poster
Applied NLP - PosterApplied NLP - Poster
Applied NLP - Poster
 
Applied CNN Poster
Applied CNN PosterApplied CNN Poster
Applied CNN Poster
 
AI Challenges Expert Need Poster
AI Challenges Expert Need PosterAI Challenges Expert Need Poster
AI Challenges Expert Need Poster
 

Recently uploaded

Criminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdfCriminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdf
Criminal IP
 
一比一原版(CU毕业证)卡尔顿大学毕业证成绩单
一比一原版(CU毕业证)卡尔顿大学毕业证成绩单一比一原版(CU毕业证)卡尔顿大学毕业证成绩单
一比一原版(CU毕业证)卡尔顿大学毕业证成绩单
yhkoc
 
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Subhajit Sahu
 
FP Growth Algorithm and its Applications
FP Growth Algorithm and its ApplicationsFP Growth Algorithm and its Applications
FP Growth Algorithm and its Applications
MaleehaSheikh2
 
【社内勉強会資料_Octo: An Open-Source Generalist Robot Policy】
【社内勉強会資料_Octo: An Open-Source Generalist Robot Policy】【社内勉強会資料_Octo: An Open-Source Generalist Robot Policy】
【社内勉強会資料_Octo: An Open-Source Generalist Robot Policy】
NABLAS株式会社
 
Ch03-Managing the Object-Oriented Information Systems Project a.pdf
Ch03-Managing the Object-Oriented Information Systems Project a.pdfCh03-Managing the Object-Oriented Information Systems Project a.pdf
Ch03-Managing the Object-Oriented Information Systems Project a.pdf
haila53
 
一比一原版(NYU毕业证)纽约大学毕业证成绩单
一比一原版(NYU毕业证)纽约大学毕业证成绩单一比一原版(NYU毕业证)纽约大学毕业证成绩单
一比一原版(NYU毕业证)纽约大学毕业证成绩单
ewymefz
 
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单
ewymefz
 
一比一原版(UVic毕业证)维多利亚大学毕业证成绩单
一比一原版(UVic毕业证)维多利亚大学毕业证成绩单一比一原版(UVic毕业证)维多利亚大学毕业证成绩单
一比一原版(UVic毕业证)维多利亚大学毕业证成绩单
ukgaet
 
Criminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdfCriminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdf
Criminal IP
 
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
John Andrews
 
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
ewymefz
 
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
nscud
 
Tabula.io Cheatsheet: automate your data workflows
Tabula.io Cheatsheet: automate your data workflowsTabula.io Cheatsheet: automate your data workflows
Tabula.io Cheatsheet: automate your data workflows
alex933524
 
1.Seydhcuxhxyxhccuuxuxyxyxmisolids 2019.pptx
1.Seydhcuxhxyxhccuuxuxyxyxmisolids 2019.pptx1.Seydhcuxhxyxhccuuxuxyxyxmisolids 2019.pptx
1.Seydhcuxhxyxhccuuxuxyxyxmisolids 2019.pptx
Tiktokethiodaily
 
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Subhajit Sahu
 
一比一原版(CBU毕业证)卡普顿大学毕业证成绩单
一比一原版(CBU毕业证)卡普顿大学毕业证成绩单一比一原版(CBU毕业证)卡普顿大学毕业证成绩单
一比一原版(CBU毕业证)卡普顿大学毕业证成绩单
nscud
 
一比一原版(YU毕业证)约克大学毕业证成绩单
一比一原版(YU毕业证)约克大学毕业证成绩单一比一原版(YU毕业证)约克大学毕业证成绩单
一比一原版(YU毕业证)约克大学毕业证成绩单
enxupq
 
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
axoqas
 
Innovative Methods in Media and Communication Research by Sebastian Kubitschk...
Innovative Methods in Media and Communication Research by Sebastian Kubitschk...Innovative Methods in Media and Communication Research by Sebastian Kubitschk...
Innovative Methods in Media and Communication Research by Sebastian Kubitschk...
correoyaya
 

Recently uploaded (20)

Criminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdfCriminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdf
 
一比一原版(CU毕业证)卡尔顿大学毕业证成绩单
一比一原版(CU毕业证)卡尔顿大学毕业证成绩单一比一原版(CU毕业证)卡尔顿大学毕业证成绩单
一比一原版(CU毕业证)卡尔顿大学毕业证成绩单
 
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
Levelwise PageRank with Loop-Based Dead End Handling Strategy : SHORT REPORT ...
 
FP Growth Algorithm and its Applications
FP Growth Algorithm and its ApplicationsFP Growth Algorithm and its Applications
FP Growth Algorithm and its Applications
 
【社内勉強会資料_Octo: An Open-Source Generalist Robot Policy】
【社内勉強会資料_Octo: An Open-Source Generalist Robot Policy】【社内勉強会資料_Octo: An Open-Source Generalist Robot Policy】
【社内勉強会資料_Octo: An Open-Source Generalist Robot Policy】
 
Ch03-Managing the Object-Oriented Information Systems Project a.pdf
Ch03-Managing the Object-Oriented Information Systems Project a.pdfCh03-Managing the Object-Oriented Information Systems Project a.pdf
Ch03-Managing the Object-Oriented Information Systems Project a.pdf
 
一比一原版(NYU毕业证)纽约大学毕业证成绩单
一比一原版(NYU毕业证)纽约大学毕业证成绩单一比一原版(NYU毕业证)纽约大学毕业证成绩单
一比一原版(NYU毕业证)纽约大学毕业证成绩单
 
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单
一比一原版(UMich毕业证)密歇根大学|安娜堡分校毕业证成绩单
 
一比一原版(UVic毕业证)维多利亚大学毕业证成绩单
一比一原版(UVic毕业证)维多利亚大学毕业证成绩单一比一原版(UVic毕业证)维多利亚大学毕业证成绩单
一比一原版(UVic毕业证)维多利亚大学毕业证成绩单
 
Criminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdfCriminal IP - Threat Hunting Webinar.pdf
Criminal IP - Threat Hunting Webinar.pdf
 
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
Chatty Kathy - UNC Bootcamp Final Project Presentation - Final Version - 5.23...
 
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
一比一原版(UPenn毕业证)宾夕法尼亚大学毕业证成绩单
 
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
一比一原版(CBU毕业证)不列颠海角大学毕业证成绩单
 
Tabula.io Cheatsheet: automate your data workflows
Tabula.io Cheatsheet: automate your data workflowsTabula.io Cheatsheet: automate your data workflows
Tabula.io Cheatsheet: automate your data workflows
 
1.Seydhcuxhxyxhccuuxuxyxyxmisolids 2019.pptx
1.Seydhcuxhxyxhccuuxuxyxyxmisolids 2019.pptx1.Seydhcuxhxyxhccuuxuxyxyxmisolids 2019.pptx
1.Seydhcuxhxyxhccuuxuxyxyxmisolids 2019.pptx
 
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
Algorithmic optimizations for Dynamic Levelwise PageRank (from STICD) : SHORT...
 
一比一原版(CBU毕业证)卡普顿大学毕业证成绩单
一比一原版(CBU毕业证)卡普顿大学毕业证成绩单一比一原版(CBU毕业证)卡普顿大学毕业证成绩单
一比一原版(CBU毕业证)卡普顿大学毕业证成绩单
 
一比一原版(YU毕业证)约克大学毕业证成绩单
一比一原版(YU毕业证)约克大学毕业证成绩单一比一原版(YU毕业证)约克大学毕业证成绩单
一比一原版(YU毕业证)约克大学毕业证成绩单
 
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
哪里卖(usq毕业证书)南昆士兰大学毕业证研究生文凭证书托福证书原版一模一样
 
Innovative Methods in Media and Communication Research by Sebastian Kubitschk...
Innovative Methods in Media and Communication Research by Sebastian Kubitschk...Innovative Methods in Media and Communication Research by Sebastian Kubitschk...
Innovative Methods in Media and Communication Research by Sebastian Kubitschk...
 

Machine Learning , Decision Tress and Novelty Detection

  • 1. Machine Learning Algorithms (MLAs) and Novelty Detection Rangaprasad Sampath (ranga.sampath@gmail.com) Madhusoodhana Chari S (madhucharis@gmail.com)
  • 2. Teaching a Child and a MLA (Machine Learning Algorithm) Orange Apple
  • 3. Responses to Unseen/Novel Objects Insights Humans may not classify Unseen/Novel objects into known classes. MLAs classify Unseen/Novel objects into the most probable class. Orange Orange (30%), Apple (70%) Apple I don’t know Orange (72%), Apple (28%) No response
  • 4. Decision Tree Classifier Decision Tree Classifier labels as a Car or Not based on the measurements of length (x2) and width (x1) of a vehicle. 4 x1 Width (m) x2 Length (m) Label 4 9 !Car 4 10 !Car 4 12 Car 3 11 Car 7 10 !Car 8 16 Car 12 18 Car 10 18 Car Training Data
  • 5. Novelty Classification Test Data that is very different from what has been seen during training i.e. Novelty, is classified into one of the known classes. 5 x1 x2 Label 4 9 !Car 4 10 !Car 4 12 Car 3 11 Car Test Data (Red dots)
  • 6. Recognition of an Unknown class Proposal: Label test data points that represent Novelty as Unknown. 6 x1 x2 Label 4 9 Unknown 4 10 Unknown 4 12 Unknown 3 11 Unknown Test Data (Red dots) Unknown Unknown Unknown Unknown
  • 7. Benefits • Avoidance – of Novel data misclassification. • Utility – in domains where it is difficult to obtain fully representative datasets for training. • Retraining – could be triggered if the number of data points classified as Unknown exceeded the number of data points classified into known classes. • Alerts – could be raised if the percentage of data that got classified as an Unknown class exceeded a set threshold (say 10 %).

Editor's Notes

  1. This is a sample Three Pictures with Captions slide ideal for including three pictures with brief, bulleted text. To Replace the Pictures on this Sample Slide (this applies to all slides in this template that contain replaceable pictures) Select the sample picture and press Delete. Click the icon inside the shape to open the Insert Picture dialog box. Navigate to the location where the picture is stored, select desired picture and click on the Insert button to fit the image proportionally within the shape. Note: Do not right-click the image to change the picture inside the picture placeholder. This will change the frame size of the picture placeholder. Instead, follow the steps outlined above. Tip: use the Crop tool to reposition a picture within a placeholder. From the Picture Tools Format tab on the ribbon, click the Crop button. Click and drag the picture within the placeholder to reposition. To scale the picture within the placeholder (while Crop is active), grab a round corner handle and drag to resize. Hold Shift key to constrain picture aspect ratio when resizing.
  2. This is a sample Three Pictures with Captions slide ideal for including three pictures with brief, bulleted text. To Replace the Pictures on this Sample Slide (this applies to all slides in this template that contain replaceable pictures) Select the sample picture and press Delete. Click the icon inside the shape to open the Insert Picture dialog box. Navigate to the location where the picture is stored, select desired picture and click on the Insert button to fit the image proportionally within the shape. Note: Do not right-click the image to change the picture inside the picture placeholder. This will change the frame size of the picture placeholder. Instead, follow the steps outlined above. Tip: use the Crop tool to reposition a picture within a placeholder. From the Picture Tools Format tab on the ribbon, click the Crop button. Click and drag the picture within the placeholder to reposition. To scale the picture within the placeholder (while Crop is active), grab a round corner handle and drag to resize. Hold Shift key to constrain picture aspect ratio when resizing.
  3. This is a sample Three Pictures with Captions slide ideal for including three pictures with brief, bulleted text. To Replace the Pictures on this Sample Slide (this applies to all slides in this template that contain replaceable pictures) Select the sample picture and press Delete. Click the icon inside the shape to open the Insert Picture dialog box. Navigate to the location where the picture is stored, select desired picture and click on the Insert button to fit the image proportionally within the shape. Note: Do not right-click the image to change the picture inside the picture placeholder. This will change the frame size of the picture placeholder. Instead, follow the steps outlined above. Tip: use the Crop tool to reposition a picture within a placeholder. From the Picture Tools Format tab on the ribbon, click the Crop button. Click and drag the picture within the placeholder to reposition. To scale the picture within the placeholder (while Crop is active), grab a round corner handle and drag to resize. Hold Shift key to constrain picture aspect ratio when resizing.