SlideShare a Scribd company logo

What should be your approach for solving ml cv problem statements

Download as PPTX, PDF
V
Vishwas N

The document provides advice and solutions for taking on a first ML/CV project. It recommends starting with the basics, focusing on relevant concepts, and not rushing the project. It also suggests identifying problems in one's surroundings to define a problem statement, and participating in competitions but not focusing on leaderboards. The document outlines a sequential approach starting with collecting and understanding data, then considering appropriate models, training methods, and faster/better training.

Internet
1 of 86
Download to read offline
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
32
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
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
What should be your approach for solving ML/CV problem statements? By : Vishwas Narayan
What should be your approach for solving ml cv problem statements
What should be your approach for solving ml cv problem statements
What should be your approach for solving ml cv problem statements
What should be your approach for solving ml cv problem statements
I dont what should be the title ● Your first ML/CV project β—‹ Thinking about a problem statement FIRST. β—‹ Things to consider for the project (frameworks,Compute,Algorithm). β—‹ Executing the project is a EGO problem. β—‹ Presenting the project and having feedback is Chaotic. β—‹ ...etc,etc ● What's next or what can it be that's the big challenge.
What should be your approach for solving ml cv problem statements
Motivation
With their blessings
What should be your approach for solving ml cv problem statements
What should be your approach for solving ml cv problem statements
What should be your approach for solving ml cv problem statements
What should be your approach for solving ml cv problem statements
What should be your approach for solving ml cv problem statements
What should be your approach for solving ml cv problem statements
What should be your approach for solving ml cv problem statements
Your first ML/CV project - Challenges ● How much should you know before you start your ML/CV project?
Your first ML/CV project - Challenges ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed?
Your first ML/CV project - Challenges ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? ● What problem statement do you choose for the project?(decide on what factors)
What should be your approach for solving ml cv problem statements
Your first ML/CV project - Challenges ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? ● What problem statement do you choose for the project?(decide on what factors) ● Lots of moving parts in the project. Where do you start?
Let me just say one thing Engineering tools are open source but engineers value is not open sourced.
What technology is nowhere related to what technology can be.
Let’s talk about solutions now!
What should be your approach for solving ml cv problem statements
What should be your approach for solving ml cv problem statements
Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? β—‹ You don’t need to know everything before you start the project. Just basic concepts should be enough to start off but dont sit with one thing only as you have to pay your bills.
Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? β—‹ You don’t need to know everything before you start the project. Just basic concepts should be enough to start off but dont sit with one thing only as you have to pay your bills. β—‹ I typically follow a 30:40:20:10 ratio (things I know : Algorithm : learning opportunities : Documentation). Make your own ratios and adjust them over time and never learn and leave things.
Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? β—‹ You cannot be a master of everything at any given time thus stay calm thus master in some things at least.
Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? β—‹ You cannot be a master of everything at any given time thus stay calm. β—‹ Focus on the basics that are still missing. Focus on the things relevant for executing the project and also try some new things.
Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? β—‹ You cannot be a master of everything at any given time. β—‹ Focus on the basics that are still missing. Focus on the things relevant for executing the project and also try some new things. β—‹ Don’t rush it out! As it gets too bad.
Like i had to even make a new slide Take the time and make sure you are understanding the things as you are going along.
Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for no reason? β—‹ You cannot be a master of everything at any given time. β—‹ Focus on the basics that are still missing. Focus on the things relevant for executing the project. β—‹ Don’t rush it out! As it gets too bad. β—‹ So, how much understanding is needed?
Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? β—‹ [...] β—‹ Don’t rush it out! As it gets too bad. β—‹ So, how much understanding is needed for you (U decide it)? β–  Just enough to convince yourself! and clients Understanding can always be iterated!(do it in a good way). My tip: Empty your mind and try doing it.
Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for all reason? And also even a little bacha knows it and convincing new people is a very bad joke.
Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for all reason? ● What problem statement do you choose for the project? β—‹ Try identifying the problems in your surroundings - places you visit, platforms you use and so on thus you get to know what industry needs.
Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for all reason? ● What problem statement do you choose for the project? β—‹ Try identifying the problems in your surroundings - places you visit, platforms you use and so on thus for industry. β—‹ Decide if the problem is ML/CV problem framing to set a realistic goal for your safety.
Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for all reason? ● What problem statement do you choose for the project? β—‹ Try identifying the problems in your surroundings - places you visit, platforms you use and so on thus for the job. β—‹ Decide if the problem is ML/CV friendly thus you don't have the situation where you don't know the solution. β—‹ Participate in Kaggle competitions, but ...
Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for all reason? ● What problem statement do you choose for the project? β—‹ [...] β—‹ Participate in Kaggle competitions, but … β–  Don’t chase the leaderboard! As it's a whole new level its β€œCompetitive Data Science”.
It's too horrible
And beware of this
Your first ML/CV project - Solutions ● How much should I know before I start my ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for all reason? ● What problem statement do you choose for the project?(learn priority) β—‹ [...] β—‹ Participate in Kaggle competitions, but … β–  Don’t chase the leaderboard! as the competitive data scientist. β–  Always think of the bigger picture.
Do that
Your first ML/CV project - Solutions ● How much should I know before I start my ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? ● What problem statement do you choose for the project? β—‹ [...] β—‹ Participate in Kaggle competitions, but … β–  [...] β–  Here’s an amazing resource to learn creative ML/CV: https://www.visualdata.io/ OR https://mlwave.com/
Executing an ML/CV project: Some thoughts you should give
Some things to consider ● I find it useful to start with a sequential/systemic flow of learning and making the project: You work for the sequential data.
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data!
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! β—‹ How do I collect data?
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data!(it's the new technology today) β—‹ How do I collect data?(Kaggle and other online resources, web scraping, manually collect. data, etc.)
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! β—‹ How do I collect data?(Kaggle and other online resources, web scraping, manually collect data, etc.) β—‹ Is there a similar dataset available?
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! β—‹ How do I collect data?(Kaggle and other online resources, web scraping, manually collect data, etc.) β—‹ Is there a similar dataset available? β—‹ How do I become one with the data? Let's say the truth. There is no alternative to knowing your data in details(visualize it). Because data fuels ML/CV. It’s absolutely important to investigate the data in hand before starting the modeling process.
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! β—‹ [...] and a lot more.
What should be your approach for solving ml cv problem statements
What should be your approach for solving ml cv problem statements
What should be your approach for solving ml cv problem statements
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling!
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider?
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider? A lot of open source and tuned models available for the researchers today.
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider? β—‹ How do I train a model?
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider? β—‹ How do I train a model? β–  How can I train my model faster? (compute and algorithm plays a key role)
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider? β—‹ How do I train a model? β–  How can I train my model faster? β–  How can I train it better?
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider? β—‹ How do I train a model? β—‹ How do I validate a model?
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider? β—‹ How do I train a model? β—‹ How do I validate a model? β—‹ How do I debug a model?
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ [...] β—‹ How do I debug a model? β–  Check out this course:the best one is here https://developers.google.com/machine-learning/testing-debugging
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! ● Does my ML model integrate well with other systems?
Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! ● Does my ML model integrate well with other systems? β—‹ A web application
Some things to consider ● I find it consistent to start with a sequential flow. ● Data! ● Modeling! ● Does my ML model integrate well with other systems? β—‹ A web application β—‹ A mobile application
Some things to consider ● I find it consistent to start with a sequential flow. ● Data! ● Modeling! ● Does my ML model integrate well with other systems? β—‹ A web application β—‹ A mobile application β—‹ A coral board β—‹ Arduino
Some things to consider ● I find it consistent to start with a sequential flow. ● Data! ● Modeling! ● Does my ML model integrate well with other systems? Possibilities are endless here! Thus be careful!
Some things to consider ● [...] ● Does my ML model integrate well with other systems? ● Think about the other components of your project too, not just ML/CV!
Presenting your project
β€œProject” out for the world ● Nothing like a structured GitHub repository. (Like a it should be readable and reusable)
β€œProject” out for the world ● Nothing like a structured GitHub repository. β—‹ Be sure to add a proper README.md including a demo of your project.
What should be your approach for solving ml cv problem statements
β€œProject” out for the world ● Nothing like a structured GitHub repository. β—‹ Be sure to add a proper README.md including a demo of your project. β—‹ A polished directory structure.(keep it simple also) https://analyticsindiamag.com/behind-the-code-meet-abhishek-thakur- worlds-first-kaggle-triple-grandmaster/
β€œProject” out for the world ● Nothing like a structured GitHub repository. ● Write out a blog on the project and be as detailed as you can be! And this brings a lot of clarity.
β€œProject” out for the world ● Nothing like a structured GitHub repository. ● Write out a blog on the project and be as detailed as you can be! β—‹ Weights and Biases Content Developers β—‹ Nanonets Writers β—‹ Medium β—‹ Kdnuggets β—‹ Etc etc Paid** writing opportunities so money will come to you when you are learning.
β€œProject” out for the world ● Nothing like a structured GitHub repository. ● Write out a blog on the project and be as detailed as you can be! ● Share your project with the communities that you can reach or else join one learn and make the best use of it.
β€œProject” out for the world ● Nothing like a structured GitHub repository. ● Write out a blog on the project and be as detailed as you can be! ● Share your project with the communities and also ask a lot of questions. β—‹ AIDL Facebook Group β—‹ Twitter β—‹ FastAI Forums β—‹ Tensorflow Hub
β€œProject” out for the world ● Nothing like a structured GitHub repository. ● Write out a blog on the project and be as detailed as you can be! ● Share your project with the communities. ● Be open to constructive feedback thus lose your ego before you hear it and also keep it in mind.
What should be your approach for solving ml cv problem statements
Some more tips ● Figure out what interests you and keep it for a long time. ML/CV is a huge literally huge!
Some more tips ● Figure out what interests you and keep it for a long time. ML/CVis a huge!(the truth) β—‹ Some interesting articles in linkedin and medium.
Some more tips ● Figure out what interests you and keep it for a long time. ML/CV is a huge field! β—‹ Some interesting articles in linkedin and medium. ● Discuss your work with like-minded people.
Some more tips ● Figure out what interests you and keep it for a long time. ML/CV is a huge field! β—‹ Some interesting articles in linkedin and medium. ● Discuss your work with like-minded people. ● Finally hustle,search for the solutions and problem statement,apply, learn, make mistakes and repeat!
If you want to further learn Please do contact me from any social media platform that I have listed here Email : vishwasnarayan2345@gmail.com Linkedin Vishwas N : shorturl.at/aquyC Twitter : shorturl.at/gios7

More Related Content

PPTX
Quantum machine learning basics 2
Krishna Kumar Sekar
Β 
PPTX
Quantum machine learning basics
Krishna Kumar Sekar
Β 
PPTX
Machine learning
Alen Varghese
Β 
PDF
From Lab to Factory: Or how to turn data into value
Peadar Coyle
Β 
PDF
The pragmatic programmer
Joel CorrΓͺa
Β 
PDF
Programming love
Andres Bedoya Tobon
Β 
PDF
How to become a data scientist
Manjunath Sindagi
Β 
PPTX
Product School - AI Funding / Trends & Product Management
Aarthi Srinivasan
Β 
Quantum machine learning basics 2
Krishna Kumar Sekar
Β 
Quantum machine learning basics
Krishna Kumar Sekar
Β 
Machine learning
Alen Varghese
Β 
From Lab to Factory: Or how to turn data into value
Peadar Coyle
Β 
The pragmatic programmer
Joel CorrΓͺa
Β 
Programming love
Andres Bedoya Tobon
Β 
How to become a data scientist
Manjunath Sindagi
Β 
Product School - AI Funding / Trends & Product Management
Aarthi Srinivasan
Β 

Similar to What should be your approach for solving ml cv problem statements (20)

PDF
What should be your approach for solving ML_CV problem statements_.pdf
Vishwas N
Β 
PPTX
From Engineering to Product Management
Janko Bazhdavela
Β 
PDF
build@mercari-week7-mark-talk
Mark Hahn
Β 
PPTX
How to Build your Career.pptx
vaideheekore
Β 
PPTX
Data Driven Business Lab Feb2019
Dries van den Enden
Β 
PDF
Cto meetup Berlin
Alexander Grosse
Β 
PPTX
Presentation for JSPM's RSCOE
RiteshMenon3
Β 
PDF
How Product Managers & Developers Deliver Value at Avvo
Danielle Martin
Β 
PDF
Startup Roles and Responsibilities + Share Structure
Ali Khajeh-Hosseini
Β 
PDF
CP vs Project - Elevate Ep. 02.pdf
preetikumara
Β 
PPTX
Employment 10 year plan Luke Blackman
Louise Reed
Β 
PDF
How to requirements inc.com
Srinivas S
Β 
PPT
Congratulations, you have been promoted to a manager role. You`ve got new pro...
Lohika_Odessa_TechTalks
Β 
PDF
Putting real time into practice - Saul Diez-Guerra
Future Insights
Β 
PDF
Ace the Tech Interviews - www.hiredintech.com
Anton Dimitrov
Β 
PDF
How i got interviews at google, facebook, and bridgewater (tech version)
Tomiwa Ademidun
Β 
PDF
How to build a successful career with your professional Certifications
Ojiugo Ajunwa
Β 
PDF
Starting your career as UX designer during pandemic
Arnold Saputra
Β 
PDF
How to play & win the product management career game
Ankur Sharma
Β 
PDF
How to become Industry ready engineers.pdf
DrNilam Choudhary
Β 
What should be your approach for solving ML_CV problem statements_.pdf
Vishwas N
Β 
From Engineering to Product Management
Janko Bazhdavela
Β 
build@mercari-week7-mark-talk
Mark Hahn
Β 
How to Build your Career.pptx
vaideheekore
Β 
Data Driven Business Lab Feb2019
Dries van den Enden
Β 
Cto meetup Berlin
Alexander Grosse
Β 
Presentation for JSPM's RSCOE
RiteshMenon3
Β 
How Product Managers & Developers Deliver Value at Avvo
Danielle Martin
Β 
Startup Roles and Responsibilities + Share Structure
Ali Khajeh-Hosseini
Β 
CP vs Project - Elevate Ep. 02.pdf
preetikumara
Β 
Employment 10 year plan Luke Blackman
Louise Reed
Β 
How to requirements inc.com
Srinivas S
Β 
Congratulations, you have been promoted to a manager role. You`ve got new pro...
Lohika_Odessa_TechTalks
Β 
Putting real time into practice - Saul Diez-Guerra
Future Insights
Β 
Ace the Tech Interviews - www.hiredintech.com
Anton Dimitrov
Β 
How i got interviews at google, facebook, and bridgewater (tech version)
Tomiwa Ademidun
Β 
How to build a successful career with your professional Certifications
Ojiugo Ajunwa
Β 
Starting your career as UX designer during pandemic
Arnold Saputra
Β 
How to play & win the product management career game
Ankur Sharma
Β 
How to become Industry ready engineers.pdf
DrNilam Choudhary
Β 
Ad

More from Vishwas N (20)

PDF
API Testing and Hacking.pdf
Vishwas N
Β 
PDF
API Hijacking.pdf
Vishwas N
Β 
PDF
Deepfence.pdf
Vishwas N
Β 
PDF
DevOps - A Purpose for an Institution.pdf
Vishwas N
Β 
PDF
API Testing and Hacking (1).pdf
Vishwas N
Β 
PDF
API Hijacking (1).pdf
Vishwas N
Β 
PDF
Dapr.pdf
Vishwas N
Β 
PDF
linkerd.pdf
Vishwas N
Β 
PDF
HoloLens.pdf
Vishwas N
Β 
PDF
Automated Governance for the DevOps Institutions.pdf
Vishwas N
Β 
PDF
Lets build with DevSecOps Culture.pdf
Vishwas N
Β 
PDF
Github Actions and Terraform.pdf
Vishwas N
Β 
PDF
KEDA.pdf
Vishwas N
Β 
PPTX
Ram bleed the hardware based approach for the hackers
Vishwas N
Β 
PPTX
Container on azure
Vishwas N
Β 
PPTX
Deeplearning and dev ops azure
Vishwas N
Β 
PPTX
Azure data lakes
Vishwas N
Β 
PPTX
Azure dev ops
Vishwas N
Β 
PPTX
Azure ai on premises with docker
Vishwas N
Β 
PPTX
Nlp for the precision medicine
Vishwas N
Β 
API Testing and Hacking.pdf
Vishwas N
Β 
API Hijacking.pdf
Vishwas N
Β 
Deepfence.pdf
Vishwas N
Β 
DevOps - A Purpose for an Institution.pdf
Vishwas N
Β 
API Testing and Hacking (1).pdf
Vishwas N
Β 
API Hijacking (1).pdf
Vishwas N
Β 
Dapr.pdf
Vishwas N
Β 
linkerd.pdf
Vishwas N
Β 
HoloLens.pdf
Vishwas N
Β 
Automated Governance for the DevOps Institutions.pdf
Vishwas N
Β 
Lets build with DevSecOps Culture.pdf
Vishwas N
Β 
Github Actions and Terraform.pdf
Vishwas N
Β 
KEDA.pdf
Vishwas N
Β 
Ram bleed the hardware based approach for the hackers
Vishwas N
Β 
Container on azure
Vishwas N
Β 
Deeplearning and dev ops azure
Vishwas N
Β 
Azure data lakes
Vishwas N
Β 
Azure dev ops
Vishwas N
Β 
Azure ai on premises with docker
Vishwas N
Β 
Nlp for the precision medicine
Vishwas N
Β 
Ad

Recently uploaded (20)

PPTX
Introduction to Information and Communication Technology
AkmadArzieAyao1
Β 
PPTX
Power Point - Lesson 3_2.pptx grad school presentation
JoyPPD
Β 
PDF
πŸ’° π”πŠπ“πˆ πŠπ„πŒπ„ππ€ππ†π€π πŠπˆππ„π‘πŸ’πƒ π‡π€π‘πˆ 𝐈𝐍𝐈 πŸπŸŽπŸπŸ“ πŸ’°
GRAB
Β 
PDF
Decoding a Decade: 10 Years of Applied CTI Discipline
Andreas Sfakianakis
Β 
PPTX
Funds Management Learning Material for Beg
NKKumar16
Β 
PPTX
June-4-Sermon-Powerpoint.pptx USE THIS FOR YOUR MOTIVATION
KuyaRogie
Β 
PPTX
522797556-Unit-2-Temperature-measurement-1-1.pptx
msar8888
Β 
PPTX
Introuction about WHO-FIC in ICD-10.pptx
naveenithkrishnan
Β 
PPT
isotopes_sddsadsaadasdasdasdasdsa1213.ppt
Kenneth James Alamillo
Β 
PPTX
SAP Ariba Sourcing PPT for learning material
NKKumar16
Β 
PDF
Vigrab.top – Online Tool for Downloading and Converting Social Media Videos a...
Vigrab Top
Β 
PPTX
Slides PPTX World Game (s) Eco Economic Epochs.pptx
Steven McGee
Β 
PDF
Paper PDF World Game (s) Great Redesign.pdf
Steven McGee
Β 
PPTX
E -tech empowerment technologies PowerPoint
kylebalatero12
Β 
PDF
RPKI Status Update, presented by Makito Lay at IDNOG 10
APNIC
Β 
PDF
Sims 4 Historia para lo sims 4 para jugar
lolpopy34
Β 
PPTX
CHE NAA, , b,mn,mblblblbljb jb jlb ,j , ,C PPT.pptx
sumodmjohn3
Β 
PDF
Cloud-Scale Log Monitoring _ Datadog.pdf
MuhammadDhomanhuriMa
Β 
PPT
tcp ip networks nd ip layering assotred slides
pakadamfdp
Β 
PDF
Best Practices for Testing and Debugging Shopify Third-Party API Integrations...
CartCoders
Β 
Introduction to Information and Communication Technology
AkmadArzieAyao1
Β 
Power Point - Lesson 3_2.pptx grad school presentation
JoyPPD
Β 
πŸ’° π”πŠπ“πˆ πŠπ„πŒπ„ππ€ππ†π€π πŠπˆππ„π‘πŸ’πƒ π‡π€π‘πˆ 𝐈𝐍𝐈 πŸπŸŽπŸπŸ“ πŸ’°
GRAB
Β 
Decoding a Decade: 10 Years of Applied CTI Discipline
Andreas Sfakianakis
Β 
Funds Management Learning Material for Beg
NKKumar16
Β 
June-4-Sermon-Powerpoint.pptx USE THIS FOR YOUR MOTIVATION
KuyaRogie
Β 
522797556-Unit-2-Temperature-measurement-1-1.pptx
msar8888
Β 
Introuction about WHO-FIC in ICD-10.pptx
naveenithkrishnan
Β 
isotopes_sddsadsaadasdasdasdasdsa1213.ppt
Kenneth James Alamillo
Β 
SAP Ariba Sourcing PPT for learning material
NKKumar16
Β 
Vigrab.top – Online Tool for Downloading and Converting Social Media Videos a...
Vigrab Top
Β 
Slides PPTX World Game (s) Eco Economic Epochs.pptx
Steven McGee
Β 
Paper PDF World Game (s) Great Redesign.pdf
Steven McGee
Β 
E -tech empowerment technologies PowerPoint
kylebalatero12
Β 
RPKI Status Update, presented by Makito Lay at IDNOG 10
APNIC
Β 
Sims 4 Historia para lo sims 4 para jugar
lolpopy34
Β 
CHE NAA, , b,mn,mblblblbljb jb jlb ,j , ,C PPT.pptx
sumodmjohn3
Β 
Cloud-Scale Log Monitoring _ Datadog.pdf
MuhammadDhomanhuriMa
Β 
tcp ip networks nd ip layering assotred slides
pakadamfdp
Β 
Best Practices for Testing and Debugging Shopify Third-Party API Integrations...
CartCoders
Β 

What should be your approach for solving ml cv problem statements

  • 1. What should be your approach for solving ML/CV problem statements? By : Vishwas Narayan
  • 6. I dont what should be the title ● Your first ML/CV project β—‹ Thinking about a problem statement FIRST. β—‹ Things to consider for the project (frameworks,Compute,Algorithm). β—‹ Executing the project is a EGO problem. β—‹ Presenting the project and having feedback is Chaotic. β—‹ ...etc,etc ● What's next or what can it be that's the big challenge.
  • 17. Your first ML/CV project - Challenges ● How much should you know before you start your ML/CV project?
  • 18. Your first ML/CV project - Challenges ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed?
  • 19. Your first ML/CV project - Challenges ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? ● What problem statement do you choose for the project?(decide on what factors)
  • 21. Your first ML/CV project - Challenges ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? ● What problem statement do you choose for the project?(decide on what factors) ● Lots of moving parts in the project. Where do you start?
  • 22. Let me just say one thing Engineering tools are open source but engineers value is not open sourced.
  • 23. What technology is nowhere related to what technology can be.
  • 27. Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? β—‹ You don’t need to know everything before you start the project. Just basic concepts should be enough to start off but dont sit with one thing only as you have to pay your bills.
  • 28. Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? β—‹ You don’t need to know everything before you start the project. Just basic concepts should be enough to start off but dont sit with one thing only as you have to pay your bills. β—‹ I typically follow a 30:40:20:10 ratio (things I know : Algorithm : learning opportunities : Documentation). Make your own ratios and adjust them over time and never learn and leave things.
  • 29. Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? β—‹ You cannot be a master of everything at any given time thus stay calm thus master in some things at least.
  • 30. Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? β—‹ You cannot be a master of everything at any given time thus stay calm. β—‹ Focus on the basics that are still missing. Focus on the things relevant for executing the project and also try some new things.
  • 31. Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? β—‹ You cannot be a master of everything at any given time. β—‹ Focus on the basics that are still missing. Focus on the things relevant for executing the project and also try some new things. β—‹ Don’t rush it out! As it gets too bad.
  • 32. Like i had to even make a new slide Take the time and make sure you are understanding the things as you are going along.
  • 33. Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for no reason? β—‹ You cannot be a master of everything at any given time. β—‹ Focus on the basics that are still missing. Focus on the things relevant for executing the project. β—‹ Don’t rush it out! As it gets too bad. β—‹ So, how much understanding is needed?
  • 34. Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? β—‹ [...] β—‹ Don’t rush it out! As it gets too bad. β—‹ So, how much understanding is needed for you (U decide it)? β–  Just enough to convince yourself! and clients Understanding can always be iterated!(do it in a good way). My tip: Empty your mind and try doing it.
  • 35. Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for all reason? And also even a little bacha knows it and convincing new people is a very bad joke.
  • 36. Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for all reason? ● What problem statement do you choose for the project? β—‹ Try identifying the problems in your surroundings - places you visit, platforms you use and so on thus you get to know what industry needs.
  • 37. Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for all reason? ● What problem statement do you choose for the project? β—‹ Try identifying the problems in your surroundings - places you visit, platforms you use and so on thus for industry. β—‹ Decide if the problem is ML/CV problem framing to set a realistic goal for your safety.
  • 38. Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for all reason? ● What problem statement do you choose for the project? β—‹ Try identifying the problems in your surroundings - places you visit, platforms you use and so on thus for the job. β—‹ Decide if the problem is ML/CV friendly thus you don't have the situation where you don't know the solution. β—‹ Participate in Kaggle competitions, but ...
  • 39. Your first ML/CV project - Solutions ● How much should you know before you start your ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for all reason? ● What problem statement do you choose for the project? β—‹ [...] β—‹ Participate in Kaggle competitions, but … β–  Don’t chase the leaderboard! As it's a whole new level its β€œCompetitive Data Science”.
  • 42. Your first ML/CV project - Solutions ● How much should I know before I start my ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed for all reason? ● What problem statement do you choose for the project?(learn priority) β—‹ [...] β—‹ Participate in Kaggle competitions, but … β–  Don’t chase the leaderboard! as the competitive data scientist. β–  Always think of the bigger picture.
  • 44. Your first ML/CV project - Solutions ● How much should I know before I start my ML/CV project? ● ML/CV is interdisciplinary. How to not get overwhelmed? ● What problem statement do you choose for the project? β—‹ [...] β—‹ Participate in Kaggle competitions, but … β–  [...] β–  Here’s an amazing resource to learn creative ML/CV: https://www.visualdata.io/ OR https://mlwave.com/
  • 45. Executing an ML/CV project: Some thoughts you should give
  • 46. Some things to consider ● I find it useful to start with a sequential/systemic flow of learning and making the project: You work for the sequential data.
  • 47. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data!
  • 48. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! β—‹ How do I collect data?
  • 49. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data!(it's the new technology today) β—‹ How do I collect data?(Kaggle and other online resources, web scraping, manually collect. data, etc.)
  • 50. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! β—‹ How do I collect data?(Kaggle and other online resources, web scraping, manually collect data, etc.) β—‹ Is there a similar dataset available?
  • 51. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! β—‹ How do I collect data?(Kaggle and other online resources, web scraping, manually collect data, etc.) β—‹ Is there a similar dataset available? β—‹ How do I become one with the data? Let's say the truth. There is no alternative to knowing your data in details(visualize it). Because data fuels ML/CV. It’s absolutely important to investigate the data in hand before starting the modeling process.
  • 52. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! β—‹ [...] and a lot more.
  • 56. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling!
  • 57. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider?
  • 58. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider? A lot of open source and tuned models available for the researchers today.
  • 59. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider? β—‹ How do I train a model?
  • 60. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider? β—‹ How do I train a model? β–  How can I train my model faster? (compute and algorithm plays a key role)
  • 61. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider? β—‹ How do I train a model? β–  How can I train my model faster? β–  How can I train it better?
  • 62. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider? β—‹ How do I train a model? β—‹ How do I validate a model?
  • 63. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ Which model should I consider? β—‹ How do I train a model? β—‹ How do I validate a model? β—‹ How do I debug a model?
  • 64. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! β—‹ [...] β—‹ How do I debug a model? β–  Check out this course:the best one is here https://developers.google.com/machine-learning/testing-debugging
  • 65. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! ● Does my ML model integrate well with other systems?
  • 66. Some things to consider ● I find it useful to start with a sequential/systemic flow for learning and making project. ● Data! ● Modeling! ● Does my ML model integrate well with other systems? β—‹ A web application
  • 67. Some things to consider ● I find it consistent to start with a sequential flow. ● Data! ● Modeling! ● Does my ML model integrate well with other systems? β—‹ A web application β—‹ A mobile application
  • 68. Some things to consider ● I find it consistent to start with a sequential flow. ● Data! ● Modeling! ● Does my ML model integrate well with other systems? β—‹ A web application β—‹ A mobile application β—‹ A coral board β—‹ Arduino
  • 69. Some things to consider ● I find it consistent to start with a sequential flow. ● Data! ● Modeling! ● Does my ML model integrate well with other systems? Possibilities are endless here! Thus be careful!
  • 70. Some things to consider ● [...] ● Does my ML model integrate well with other systems? ● Think about the other components of your project too, not just ML/CV!
  • 72. β€œProject” out for the world ● Nothing like a structured GitHub repository. (Like a it should be readable and reusable)
  • 73. β€œProject” out for the world ● Nothing like a structured GitHub repository. β—‹ Be sure to add a proper README.md including a demo of your project.
  • 75. β€œProject” out for the world ● Nothing like a structured GitHub repository. β—‹ Be sure to add a proper README.md including a demo of your project. β—‹ A polished directory structure.(keep it simple also) https://analyticsindiamag.com/behind-the-code-meet-abhishek-thakur- worlds-first-kaggle-triple-grandmaster/
  • 76. β€œProject” out for the world ● Nothing like a structured GitHub repository. ● Write out a blog on the project and be as detailed as you can be! And this brings a lot of clarity.
  • 77. β€œProject” out for the world ● Nothing like a structured GitHub repository. ● Write out a blog on the project and be as detailed as you can be! β—‹ Weights and Biases Content Developers β—‹ Nanonets Writers β—‹ Medium β—‹ Kdnuggets β—‹ Etc etc Paid** writing opportunities so money will come to you when you are learning.
  • 78. β€œProject” out for the world ● Nothing like a structured GitHub repository. ● Write out a blog on the project and be as detailed as you can be! ● Share your project with the communities that you can reach or else join one learn and make the best use of it.
  • 79. β€œProject” out for the world ● Nothing like a structured GitHub repository. ● Write out a blog on the project and be as detailed as you can be! ● Share your project with the communities and also ask a lot of questions. β—‹ AIDL Facebook Group β—‹ Twitter β—‹ FastAI Forums β—‹ Tensorflow Hub
  • 80. β€œProject” out for the world ● Nothing like a structured GitHub repository. ● Write out a blog on the project and be as detailed as you can be! ● Share your project with the communities. ● Be open to constructive feedback thus lose your ego before you hear it and also keep it in mind.
  • 82. Some more tips ● Figure out what interests you and keep it for a long time. ML/CV is a huge literally huge!
  • 83. Some more tips ● Figure out what interests you and keep it for a long time. ML/CVis a huge!(the truth) β—‹ Some interesting articles in linkedin and medium.
  • 84. Some more tips ● Figure out what interests you and keep it for a long time. ML/CV is a huge field! β—‹ Some interesting articles in linkedin and medium. ● Discuss your work with like-minded people.
  • 85. Some more tips ● Figure out what interests you and keep it for a long time. ML/CV is a huge field! β—‹ Some interesting articles in linkedin and medium. ● Discuss your work with like-minded people. ● Finally hustle,search for the solutions and problem statement,apply, learn, make mistakes and repeat!
  • 86. If you want to further learn Please do contact me from any social media platform that I have listed here Email : vishwasnarayan2345@gmail.com Linkedin Vishwas N : shorturl.at/aquyC Twitter : shorturl.at/gios7