Machine learning application (Kartal)

#یادگیری عمیق
یادگیری ماشین
#بیومدیکال
یک سوال زیبا و با یک پاسخ عالی
#سوال
A goal of mine is to be able to work in a job that is focused on research in Deep Learning. This seems like a difficult task in that my background is primarily in biomedical sciences with no formal computer science training. Do you think it is necessary to have a formal computer science education to contribute to Deep Learning research teams? Also what are your favorite 3 papers on Deep Learning (or the most recent 3 papers you've read)?

#پاسخ از دانشجو Andrew NG
There are many ways to contribute to Machine Learning research (or industry) teams. One is bringing knowledge from your field (BioMed Sciences) in addition to your ML training. For instance, doctors with ML training contribute significantly to ML teams to solve problems in Radiology, Cardiology and Palliative Care among others.

You can also contribute to a ML team outside your area of expertise as a Data Scientist or Machine Learning Engineer. In terms of what training would be required, I'd suggest to check out the list from our Machine Learning Engineer Career Program ( https://www.deeplearning.ai/careers/). This is quite extensive and you might not need to master everything perfectly to contribute in a ML team, but it will give you an idea:

Deep learning. You should be able to understand and apply major deep learning methods, including neural network training, regularization, optimization methods (gradient descent, Adam), and be familiar with major neural network architecture types such as Convolutional Networks, RNN/LSTM. Completion of the deeplearning.ai specialization is sufficient to meet this criteria.
Machine learning. You should be able to understand and apply major machine learning methods, such as logistic regression, SVM, Decision Trees, Principal Component Analysis and K-means. Completion of Andrew Ng’s Machine Learning course on Coursera is sufficient to meet this criteria.
Implementation. You should have prior experience taking a dataset, cleaning it if necessary, and applying a learning algorithm to it to get a result. You should be able to implement a learning algorithm “from scratch” using a framework such as Tensorflow, Pytorch, Caffe, etc.
General coding. You should be able to code non-trivial functions in object oriented programming, such as popular sorting or search algorithms. You should know how to use your terminal, and work with version control systems (Git). Software engineering experience such as working with relational databases, APIs, and building the back-end of web or mobile applications is helpful but is not required.
Mathematics (including probabilities and statistics.) You should be able to use mathematical notations and linear algebra (matrix/vector operations, dot products, etc.), and understand basic probability theory (distributions, independence, density functions, etc.) as well as statistics (mean, variance, median, quantiles, co-variance, etc.)
I've tried to learn more about DL applied to Point Clouds recently, so the 3 most recent papers I've read are:

PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation (Qi, Su et al.)
PointCNN (Li et al.)
Large-scale Point Cloud Semantic Segmentation with Superpoint Graphs (Landrieu & Simonovsky)
3 papers I like a lot:

Neural Machine Translation by Jointly Learning to Align and Translate (Bahdanau et al.) if you'd like to learn more about Machine Translation.
FaceNet: A Unified Embedding for Face Recognition and Clustering (Schroff et al.) if you'd like to learn more about Face Recognition.
YOLO9000: Better, Faster, Stronger (Redmon et al.) if you'd like to learn more about Object Detection.
Hope it helps,

منبع:
https://www.deeplearning.ai/forums/community/general-forum/meet-the-deeplearning-ai-team/

با کاناله

https://xn--r1a.website/Machinelearning_Kartal/1468

نیز همراه باشید 🌿🌿🌿💭💭💭🌹🌹🌹

www.deeplearning.ai

Careers - DeepLearning.AI

Join the DeepLearning.AI team working to make AI education accessible to the entire world!

1.29K viewsedited 13:23