In conventional neural networks, the amount of computation used is proportional to the size of the inputs, instead of the complexity of the content of the data being processed. However, the time needed to process input data is a function of more than just the size of the inputs. Common input data for neural architectures also have an inherent complexity that is independent of the input size. Conventional neural architectures do not adjust their computational budget based on the complexity of the data they are processing, or arguably, such adaptation is done manually by the machine learning practitioner. In this work, we, therefore, introduce a differentiable parameter-free Adaptive Token Sampler (ATS) module, which can be plugged into any existing vision transformer architecture. ATS empowers vision transformers by scoring and adaptively sampling significant tokens. As a result, the number of tokens is not constant anymore and varies for each input image. By integrating ATS as an additional layer within the current transformer blocks, we can convert them into much more efficient vision transformers with an adaptive number of tokens. Since ATS is a parameter-free module, it can be added to the off the-shelf pre-trained vision transformers as a plug-and-play module, thus reducing their GFLOPs without any additional training. Moreover, due to its differentiable design, one can also train a vision transformer equipped with ATS. We evaluate the efficiency of our module in both image and video classification tasks by adding it to multiple SOTA vision transformers. Our proposed module improves the SOTA by reducing their computational costs (GFLOPs) by 2X.

Very interesting idea on how to move 'common sense' in AI forward. It is always great to explore different ideas and directions and have different perspectives to increase chances of success and advancement.
https://www.technologyreview.com/2022/06/24/1054817/yann-lecun-bold-new-vision-future-ai-deep-learning-meta/

#ai #ml #dl #artificialintelligence #machinelearning #deeplearning

بهترین منابع ابزارهای هوش مصنوعی

The best Stanford, CMU, and MIT courses to build a career in AI


📚 Stanford University
🔹 CS229 - Machine Learning by Andrew Ng:

🔹 CS230 - Deep Learning by Andrew Ng

🔹 CS231n - Convolutional Neural Networks for Visual Recognition by Fei-Fei Li and Andrej Karpathy

🔹 CS224n - Natural Language Processing with Deep Learning by Christopher Manning

🔹 CS25 - Transformers United

📚 Massachusetts Institute of Technology

🔹 6.S191 - Introduction to Deep Learning by Alexander Amini and Ava Soleimany

🔹 6.S094 - Deep Learning by Lex Fridman


🔹 6.S192 - Deep Learning for Art, Aesthetics, and Creativity by Ali Jahanian

📚 Carnegie Mellon University

🔹 CS/LTI 11-777 Multimodal Machine Learning by Louis-Philippe Morency:

📚 University College London

🔹 COMP M050 Reinforcement Learning by David Silver

#فیلم #منابع #الگوریتمها #هوش_مصنوعی #پردازش_زبان_طبیعی #یادگیری_ماشین #یادگیری_عمیق #یادگیری_تقویتی

❇️ @AI_Python

ویدئو جذاب از جلسه مصاحبه و نوع نگاه مصاحبه گر به مصاحبه کننده
#متفرقه

🔰 @AI_Python

Learning Protein Representations via Complete 3D Graph Networks

DIG: Dive into Graphs is a turnkey library for graph deep learning research.

Github: https://github.com/divelab/DIG

Paper: https://arxiv.org/abs/2207.12600v1

Tutorials: https://diveintographs.readthedocs.io/en/latest/tutorials/graphdf.html

Documentation: https://diveintographs.readthedocs.io/

Benchmarks: https://github.com/divelab/DIG/tree/dig-stable/benchmarks

Dataset: https://paperswithcode.com/dataset/atom3d

Deep Deformable 3D Caricature with Learned Shape Control (DD3C)

Github: https://github.com/ycjungsubhuman/deepdeformable3dcaricatures

Paper: https://arxiv.org/abs/2207.14593v1

Project: https://ycjungsubhuman.github.io/DeepDeformable3DCaricatures

Dataset: https://paperswithcode.com/dataset/facewarehouse

Video: https://youtu.be/WLMPEaK6E4M

10 سایت برتر برای یافتن شغل در سال 2022:
(Top 10 sites for your career in 2022)
1) Linkedin
2) Indeed
3) Naukri
4) Monster
5) JobBait
6) Careercloud
7) Dice
8) CareerBuilder
9) Jibberjobber
10) Glassdoor

10 مهارت فنی مورد تقاضا در سال 2022:
(Top 10 Teach skills in demand in 2022)

1) Machine Learning
2) Mobile Development
3) SEO/SEM Marketing
4) Data Visualization
5) Data Engineering
6) UI/UX Design
7) Cyber Security
8) Cloud Computing/AWS
9) Blockchain
10) IOT

10 سایت برای آموزش آنلاین رایگان در سال 2022:
(Top 10 sites for free online education in 2022)

1) Coursera
2) edX
3) Khan Academy
4) Udemy
5) iTunesU Free Courses
6) MIT OpenCourseWare
7) Stanford Online
8) Codecademy
9) ict iitr
10) ict iitk


10 سایت برای بررسی رایگان رزومه در سال 2022:
(Top 10 sites to review your resume for free in 2022)

1) Zety Resume Builder
2) Resumonk
3) Resume dot com
4) VisualCV
5) Cvmaker
6) ResumUP
7) Resume Genius
8) Resume builder
9) Resume Baking
10) Enhance

10 سایت برای آماده سازی مصاحبه در سال 2022:
(Top 10 sites for interview Preparation in 2022)

1) Ambitionbox
2) AceThelnterview
3) Geeksforgeeks
4) Leetcode
5) Gainlo
6) Careercup
7) Codercareer
8) InterviewUp
9) InterviewBest
10) Indiabix


@BIMining

چرا برنامه نویسی؟
* حتما ببینید و به دیگران نشان بدهید که چرا برنامه نویس هستید.
باتشکر از:@DaysGone
❇️ @ai_python

ده ایده برتر در آمار که به انقلابی در هوش مصنوعی منجر شد
https://news.columbia.edu/news/top-10-ideas-statistics-ai

❗️ مطلب داغ روز

مقایسه تفاوت « شبکه های عصبی و درخت تصمیم»

#NeuralNetworks #decisiontree
#شبکه_عصبی #درخت_تصمیم #آموزش #منابع


❇️ @AI_Python
🗣 @AI_Python_arXiv
✴️ @AI_Python_EN

چگونه اولین شبکه عصبی مصنوعی را در پایتون درست کنیم ؟؟

#مقاله #کتاب #آموزش #شبکه_عصبی_مصنوعی #منابع #فیلم #پایتون

❇️ @AI_Python
🗣 @AI_Python_arXiv
✴️ @AI_Python_EN

انتشار دوره کلاس آموزشی دانشگاه Purdue
Machine Learning For Cyber Security

#یادگیری_ماشین #منابع #کلاس_آموزشی

🔰 @AI_Python

Natural Language Processing from Stanford University: Distilled Notes

👉🏼 http://nlp.aman.ai
#منابع

🔰 @AI_Python

The Complete Collection of Data Science Projects

https://www.kdnuggets.com/2022/08/complete-collection-data-science-projects-part-1.html

#معرفی سایت

تو این سایت می‌تونید یه کلمه یا collocation تو انگلیسی را سرچ کنید، و براتون توی روزنامه ها و سایتهای معروف میگرده و هر جا این عبارت استفاده شده را میاره،
یه کاربرد عالیش برای پیدا کرد مثالهای واقعی استفاده از یک اصطلاح یا لغته:

https://ludwig.guru/