Microsoft Cognitive Toolkit (Formerly CNTK) Pros and Cons

An amazing tool for the machine learning. With help from Github tutorials we can easily setup the Cognitive tool on any of the windows or Linux machines. Being a .NET developer working with Cognitive toolkit is very easy as it supports c#. When we started using this toolkit, we have deployed this on Ubuntu machine and did the POC Review collected by and hosted on G2.com.

No major flaws found with this toolkit may be because we did not find any other alternative. But one thing we noticed was the core api's are created in c++, few times it would be difficult to debug the application in Python.

Microsoft is working hard to create wrappers but a time consuming process. Review collected by and hosted on G2.com.

I feel CNTK is one of the most widely known machine learning frameworks in the market it features great compatibility and effective use of computational resources. Review collected by and hosted on G2.com.

Its complicated to figure out how to command exactly what i want the machine to learn. I know but machines come up with multiple outcomes. it gets demotivating over time.

The UI/UX can be made better than what is is right now. probably adopt or create something similar to tableu's features. Review collected by and hosted on G2.com.

Easily scalable across business. I have used Face API for detection emotion which I truly enjoyed in my workplace. I highly recommend this Toolkit because of its ease of use and the speed for processing data. Review collected by and hosted on G2.com.

Some of the subscription keys have not worked in past but I have not had a problem in quite some time now. Missing out on full Microsoft customer support experience because of the community of open source. Review collected by and hosted on G2.com.

This is one of the easier to get started tool for machine learning with minimal coding experience. The product also has good interfaces that can be leveraged for someone trying to get started with machine learning and build models. Review collected by and hosted on G2.com.

To be honest, the one thing that can perhaps be improved is the content that is available in the machine learning model gallery - when I was trying this product, there were more images available in the gallery than other models. Review collected by and hosted on G2.com.

My favorite part about the MC toolkit is that it helps you understand python image recognition. It is very easy to install and initialize the library and naive person who is used to python can easily do that.The website describes it as a harness and I would agree with that. Review collected by and hosted on G2.com.

The only part that I dislike is that it can be hard to understand at first. I suggest watching a demo before just launching into the program Review collected by and hosted on G2.com.

I like how simple it was to incorporate this software after being in the midst of changing software. I was not employed when the previous software was being used, however I love the simplicity and versatility of this software. Review collected by and hosted on G2.com.

Honestly there was nothing that I disliked about this software, but that might have been because I did not have anything to compare this software to. Since I was not employed at this company when the previous software was being used. Review collected by and hosted on G2.com.

strong, modular and easily scaleable across business requirement Review collected by and hosted on G2.com.

Was not as easy to gain support among our end clients for this product (potentially due to our target audience). Pricing and support issues initially. Review collected by and hosted on G2.com.

Microsoft Cognitive Toolkit or CNTK is the best toolkit available for python for image recognition.Also it can be used with many other languages that you use.It gives you power to use the intelligence of very large datasets which helps you get the variety of data.Also it give you commercial-grade quality and support as stated on its website.You can get very good accuracy and that also with less time while training on its gpu which reduces your cpu usage.It also has many other features which you can check it out on the CNTK website.Support document is also good.It has everything stated which can be understood very easily right from installation to creating your own classifier and everything in between.It is a good API and must try once. Review collected by and hosted on G2.com.

I have nothing to dislike about Microsoft's CNTK. Review collected by and hosted on G2.com.

Microsoft cognitive toolkit or CNTK is basically API which is available for python and other platforms.It is very easy to install and initialize the library and naive person who is used to python can easily do that.It has many great features like you can use the pretrained model for cancer classification using logistic regression.You just have to pass the test data to the model and you will get the results.Also you can change the weights for the algorithm if you wish to do so.

Now regarding training your own model.Suppose if you wish to train your custom logistic regression classifier than you can use your own data set.Create your own test data using cross_validation function of the api and train the data.It is so simple and it can be very useful for training custom models.

Summing up CNTK is the library that any ML engineer can dream of because it has so many pretrained models like MNIST dataset which reduces the efforts. Review collected by and hosted on G2.com.

I have nothing to dislike about CNTK because it is providing so much functionality and also it is opensource. Review collected by and hosted on G2.com.

I like as primary thing that the product opensource, friendly use and with the integration of new technologies (cognitive analysis even with pictures) Review collected by and hosted on G2.com.

I dislike that this product is available only for Windows (latest versions) and not for Unix distributions Review collected by and hosted on G2.com.