Deep Learning

Deep Learning is a creation of machines which use techniques inspired by the human brains ability to learn recently we simply didn’t have enough data and processing power to train a machine to learn deep learning neural networks learn many levels of abstraction they range from simple concepts to complex ones this is what puts the deep in deep learning.

Keep one thing in mind, deep learning is not just about piling a bunch of layers of neurons and training them, it’s about the fascinating backgrounds that come into knowledge when you consider compositionality that the world around you is built of complicated features which are in turn built out of smaller easier features that are useful where you are doing machine learning in general.

To interpret what deep learning is, it’s first essential to distinguish it from the other branch of knowledge within the field of AI. One thing to keep in mind is that it is highly time-consuming.

Why everyone wants to be in the deep learning game?

In 2011 Andrew Ng Stanford’s CS professor founded Google’s GOOGLE Brain project, which formed a neural network trained with deep learning algorithms, which excellently proved capable of accepting high-level concepts such as cats.

Deep networks can learn features in an unsupervised manner. A lot of classical work in machine learning for practical applications  (such as speech recognition, image classifications) involved handcrafting features for the particular applications. Deep learning takes features engineering out of the picture.

With enough data and a good network architecture ( there are several heuristics one can use crafting good architectures for a particular problem), neutrons in a deep neural network can discover abstract features the deeper you will go in the network, the more abstract the features will be. So for many other applications, you can just drop in your deep network and let the network discover features by itself.

 The neural network not only learns how to learn these features but also knows how to combine them well. This is because it knows how crucial definite features are compared to others for the categorization task at hand. This distributed feature representation is therefore very powerful (in some sense, it has more degrees of freedom and therefore it can approximate more complex functions well) compared to other learning representations.

Again, in fiat to discover such a representation, you need a lot of data. However, as the world beget more and more data each day, it is only sensible to use technologies which can discover features from them in a machine-driven fashion.

And with past betterments in GPU technology courtesy to gamers out there, a lot of matrix computations (which are computational bottlenecks) can be done very efficiently in parallel. Therefore these were the some of the reasons why deep leaning is important to humankind.

Let us know what do you think about deep learning and why is it important? You can give suggestions by Email us: This email address is being protected from spambots. You need JavaScript enabled to view it.

 

Image credit: gigaom

Artificial intelligence has been really transforming the world of medicine. It has introduced tons of new processes, methods which has helped the doctors to diagnose a disease quicker, predict the future symptoms and cure it at a faster pace. Artificial intelligence has been around in this industry for a fair bit of time now, deep-learning is the new upcoming technology that has made things even more interesting.

With the help of deep-learning, researchers are now able to analyse the medical data for better treatment, it has enhanced the doctor’s capability to analyse the medical images etc.

Importance of deep learning: -

  1. Mining data for better and faster treatment: -

GPU- accelerated deep-learning technology can be used in order to study the patient’s records and symptoms over a long period of time and have a comparison with the larger population in order to have a better treatment process.

  1. Better Diagnosis: -

The diagnosis of patients medical reports and results such as X-ray’s, CT scans, MRI’s etc. are performed at a faster and a more accurate basis with the help of this technology which used to be a lengthy and time-consuming process earlier. Researchers of the imperial college of London have devoted a ton of time in their research in this sector.

  1. Personalized medicines: -

At the current pace of development, machine learning will be soon able to study the genomics (the study of the genetics of a patient leading to the disease) of the patient and recommend medicines known as the ‘precision’ medicine which is personalized for each patient for better and faster treatment and cure.

  1. Help to the blind: -

There are tons of people in the world who are at present partially or completely blind and face issues in performing normal day-to-day functions such as crossing roads, reading labels etc. Deep-learning technology such as wearable devices etc. are in development to help such issues.

  1. Robotic Surgery: -

Recently, there has been a lot of news on the possibility and testing of robots performing surgery on dummy limbs and vessels. ‘Da Vinci’ the robot is currently being programmed and developed in order to perform functions same as the surgeons.

Conclusion: -

Although these all opportunities are welcomed with open arms in this industry, their implementation has various challenges too. Some of them are listed below: -

  1. Robots as doctors- No doubt use of this technology and eased and quickened the medical processes leading to better efficiency. But it will never be able to replace the doctors. Doctors are much more than diagnosing and medicine-recommending figures. They study a whole bunch of stuff that the robots are currently incapable of studying.
  2. Security issues- There are a ton of security and information privacy guidelines laid down by the HIPAA which restricts the consumers to share the medical reports or test results with other medical institutions with the help of an application etc. rather than physically visiting the clinic. This restricts the use of this technology to quite an extent.

Lets understand what is deep learning.As per wikipedia info "Deep learning (also known as deep structured learning or hierarchical learning) is part of a broader family of machine learning methods based on learning data representations, as opposed to task-specific algorithms. Learning can be supervised, semi-supervised or unsupervised".

What are the real world examples and benefits of deeplearning? Below we have done some research and putting around 10 examples of deeplearning usecases.

1.Image Recognition

 - Facial Recognition

 -Image Search  ( Social Media )

-Machine Vision ( Automotive & Avaiation )

 2.Speech / Voice Recognition

- Voice search

- Sentiment analysis

-Fraud detection ( Financial firms and credit cards)

3.Text based applications

- Sentiment Analysis ( CRM,Social media,Brand Reputation Management)

-Augmented search ( Finance )

-Threat Detection ( Social media, governments )

4.Time series

-Log Analysis

-ERP

-Predictive analytics

- Recommender systems

5.Bio Informatics

6.Drug Discovery

7.Video  -

-Motion Detection ( Gaming,UI / UX)

-Real time Threat Detection ( security and airports

8.Natural Language processing

It is used by different companies in many industries, especially for negative sampling, word embedding, sentiment analysis, spoken language understanding, machine translation, contextual entity linking, and writing style recognition.

9.Customer Relationship Management

- Chat bots

-Virtual Private Assistants

10.Recommendation systems.

- Product recommendations

-Predicting user behaviours across ecommerce online shopping

Another Google use case is SmartReply . It automatically generates e-mail responses (wishing for the evolved version of this one doing our business on behalf of us) .Google Translate, which supports over 100 languages translating from one to another.

The evolved versions can be used for many cases like translating jargon legalese in contracts into plain language.Text Summarization

More use cases are evolving and 2018 year will be a remarkable for AI and machine learning,deep learning startups

 

#AI #deeplearning #machinelearning @AIMLMarketPlace

© copyright 2017 www.aimlmarketplace.com. All Rights Reserved.

A Product of HunterTech Ventures