Introduction to Deep Learning: Machine Learning vs Deep Learning

deep learning and machine learning both
offer ways to train models and classify data this video compares the two and it
offers ways to help you decide which one to use let’s start by discussing the
classic example of cats versus dogs now in this picture
do you see a cat or a dog how are you able to answer that chances are you’ve
seen many cats and dogs over time and so you’ve learned how to identify them this
is essentially what we’re trying to get a computer to do learn from and
recognize examples also keep in mind that sometimes even humans can get
identification wrong so we might expect a computer to make similar errors to
have a computer do classification using a standard machine learning approach
we’d manually select the relevant features of an image such as edges or
corners in order to train the machine learning model the model then references
those features when analyzing and classifying new objects this is an
example of object recognition however these techniques can also be
used for scene recognition and object detection when solving a machine
learning problem you follow a specific workflow you start with an image and
then you extract relevant features from it then you create a model that
describes or predicts the object on the other hand with deep learning you skip
the manual step of extracting features from images instead you feed images
directly into the deep learning algorithm which then predicts the object
so deep learning is a subtype of machine learning it deals directly with images
and is often more complex for the rest of the video when I mention machine
learning I mean anything not in the deep learning category when choosing between
machine learning and deep learning you should ask yourself whether you have a
high-performance GPU and lots of label data if you don’t have either of these
things you’ll have better luck using machine learning over deep learning this
is because deep learning is generally more complex so you’ll need at least a
few thousand images to get reliable results you’ll also need a high
performance GPU so the model spends less time analyzing those images if you
choose machine learning you have the option to train your model in
many different classifiers you may also know which features to extract that will
produce the best results plus with machine learning you have the
flexibility to choose a combination of approaches use different classifiers and
features to see which arrangement works best for your data you can use MATLAB to
try these combinations quickly also keep in mind that if you are looking to do
things like base detection you can use out-of-the-box MATLAB examples as I
mentioned before you need less data with machine learning than with deep learning
and you can get to a trained model faster too however deep learning has
become very popular recently because it is highly accurate you don’t have to
understand which features are the best representation of the object
these are learned for you but in a deep learning model you need a large amount
of data which means the model can take a long time to train you are also
responsible for many of the parameters and because the model is a black box if
something isn’t working correctly it may be hard to debug so in summary the
choice between machine learning and deep learning depends on your data and the
problem you’re trying to solve MATLAB can help you with both of these
techniques either separately or as a combined approach to find out more visit
mathworks comm slash deep learning you

42 thoughts on “Introduction to Deep Learning: Machine Learning vs Deep Learning

  • コンピューターが思考をし、最適解を導くのは、コンピューターが人間の頭脳を超えていない証拠。近い未来のAIは気付きを身に着けるだろう。これこそが、人類が長年求めていた、人間の代替えである。コンピューターが気付きを得られれば、科学は飛躍的に進歩し人類の滅亡はより早くなるに違いない。いずれにしろ人類はその存在そのものが破滅へ向かおうとしている生き物である。

  • This is only the beginning. When computers can start programming themselves and heuristically and recursively, and iteratively learn it will be at thousands of times faster than we can.

  • i found a much faster way to do machine learning than neural net. one that is shit stupid, and don't do more work than is needed to make out the features of an image and by combineing two input of information lets say picture and sound i can make the computer selflearn unsupervised. my algorithm was first an idea to make a selfdriving car that uses calculations to keep itself on the road on a desired side by modyfing its data on the fly without being able to remember anything. the method i describe can be used to create memories storied in arrays and married to works as learning. its like creating a single celled life form that can learn. the trick behind my method is to store the target image inside a numerical array, then render the array on the screen then create diagonal rays sweep left and right in the array at many angles to detect when sameness of pixels changes to non-sameness and use that to determine the distance traced on each side left and right at many angles and then store these numbers inside an array the represent the image, then do this with let say sound then combine sound and image array as a memory, so that when it recalls let say by listening to the same sound, it will show the array example of the image it extracted the information from as corresponding to the sound. so in my method i use dicrapencies in pixel colors and counting distance by how many pixels that repeat the same type until it reach a difference in the colors to measure the distances from left to right of the object its recognizing so that it can draw a crude copy of the outline of the target image and store the numbers in an array serving as a memory. if you mess up the colors lets say on a dress of a person, my computer program will not be able to recognize you outline correctly. the advantage with my method is that it can store hundres of patterns in arrays of only 8 numbers for each line therby taking up very little numbers of bytes. my idea is to create a process that cheats, takes little space and is fast. the neural net method on the other hand is slow takes a lot of space do to many trial and error operations and need a lot of computer power, not suitable for cheap microcontroller. neural net is a great, its fantatic, its just the its to power demanding and mostly for rich people or companies. i love to see progress in neural networks and perhaps even a evolution engine that mutates neural networks by natural selaction. my approach was to find a way to code a lot of information in a impossible small space, and there was no room for such a slow and heavy system as a neural net on these chips. the next evolution of computers algoritms must be to outsmart the ingenuity of nature by creating stuff on small space using little resource to do a massive job far ahead of a natural system, and not one that use even more resources and power than the natural system. its supposed to be that human intelligence was creating a artificial intelligence that was smarter than what the little molecular machines could do by combining random with order chemical reactions changing their equalibrium on a resource based computational basis. we supposed to create algoritms that outperforms the need of resources that the natural system uses by doing the algoritms smarters than natural process by using the combined intelligence of the brain to beat the molecular machinery and not creating a system that underperforms nature. we need to create a intelligece equal to the cell level that is smarter than the cell in order to make a network of this to outperform the brain. its not the complexity of the human brain that makes intelligence but the way the system is wired and the way the algoritms works than in a multiplied sense define the whole. maybe the genius programmer do exist, maybe there is many of them, they just have not gotten far enough in advances to acomplish the impossible. the collective intelligence idea is definetly an efficient one. i think both senarios works equally well but with different rules in the way of making it. one method is to do evolution with one guy or god and let the program evolve its own advancements, the other is to use humans to collectivly andvace a program by using human intelligence to solve all the problems. i think if the universe was created by one programmer, then it must have been done by a evolution engine where godly creation and evlotion is one and the same. the progrmmer create the evolution program that drives itself based on the rule set made by the programmer.

  • Deep leaning algorithms are also written by humans then why is it called "Black Box"? Can't we look into it and see how the computer solved it?

  • Does this seem ridiculous to anyone else? Deep learning is a subset of machine learning as far as i know. 'When you are doing deep learning you stop designing features and you need more data'…. uhhh ya…??? 'With deep learning you skip the step of manually extracting features'… uhhh how so? just for image classification or in general? I thought you still need to feed features to networks… do I have a major misunderstanding or is this silly?

  • Machine Learning = Supervised Classification; Deep Learning = Unsupervised Classification… in Remote Sensing.

  • Need your suggestion.
    I am new in MachineLearning/ DeepLearning and working on a project which has a large number of videos (Its requirement is to process all videos, detect objects, extract data from those and perform analysis).
    According to you, what should I start to learn first? Deep Learning or Machine learning.
    If Deep Learning is the subset of Machine learning, then will it be ok to go with Machine learning. I mean learning Machine learning will cover complete topics of Deep Learning or not.

  • Deep learning models are widely used in different fields due to its capability to handle large and complex datasets and produce the desired results with more accuracy at a greater speed. In Deep learning models, features are selected automatically through the iterative process wherein the model learns the features by going deep into the dataset and selects the features to be modeled. In the traditional models the features of the dataset needs to be specified in advance. The Deep Learning algorithms are derived from Artificial Neural Network concepts and it is a part of broader Machine Learning Models.

    This book intends to provide an overview of Deep Learning models, its application in the areas of image recognition & classification, sentiment analysis, natural language processing, stock market prediction using R statistical software package, an open source software package.

    The book also includes an introduction to python software package which is also open source software for the benefit of the users.

    This books is a second book in series after the author’s first book- Machine Learning: An Overview with the Help of R Software


    International Journal of Statistics and Medical Informatics

    Amazon Link

  • The explaination is prety informative. Is there any positive ways fpr the clues to develop and debug the predictions for the requirements more effectively. Thanks for information once again

  • This is amazing!!! Such a helpful video!! Thank you so much :)))
    This video was really helpful to me. Thank you!

  • I want to hit the like button, but the stupid machine learning is going to notice this and start recommending me other MATLAB videos that I am not interested in. Therefore, while I find this video very useful, I will not hit the like button.

Leave a Reply

Your email address will not be published. Required fields are marked *