TensorFlow is an open-source machine learning software library for numerical computation using data flow graphs. It was originally developed by Google.
N/A
Teradata Vantage
Score 8.1 out of 10
N/A
Teradata Vantage is presented as a modern analytics cloud platform that unifies everything—data lakes, data warehouses, analytics, and new data sources and types. Supports hybrid multi-cloud environments and priced for flexibility, Vantage delivers unlimited intelligence to build the future of business.
Users can deploy Vantage on public clouds (such as AWS, Azure, and GCP), hybrid multi-cloud environments, on-premises with Teradata IntelliFlex, or on commodity hardware with VMware.
TensorFlow is great for most deep learning purposes. This is especially true in two domains: 1. Computer vision: image classification, object detection and image generation via generative adversarial networks 2. Natural language processing: text classification and generation. The good community support often means that a lot of off-the-shelf models can be used to prove a concept or test an idea quickly. That, and Google's promotion of Colab means that ideas can be shared quite freely. Training, visualizing and debugging models is very easy in TensorFlow, compared to other platforms (especially the good old Caffe days). In terms of productionizing, it's a bit of a mixed bag. In our case, most of our feature building is performed via Apache Spark. This means having to convert Parquet (columnar optimized) files to a TensorFlow friendly format i.e., protobufs. The lack of good JVM bindings mean that our projects end up being a mix of Python and Scala. This makes it hard to reuse some of the tooling and support we wrote in Scala. This is where MXNet shines better (though its Scala API could do with more work).
Teradata Vantage is well suited for large scale ETL pipelines like the ones we developed for anti money laundering risk matrices. It handles heavy joins, aggregations, and transformations on transactional data efficiently. We generate alert variables, adjust for inflation, and monitor establishments monthly with it, all integrated with Python and Control-M for a centralised automation across the company. For less appropriate, I would say that heavy resource demands might slow down experimentation for iterative work.
Theano is perhaps a bit faster and eats up less memory than TensorFlow on a given GPU, perhaps due to element-wise ops. Tensorflow wins for multi-GPU and “compilation” time.
Teradata is an excellent option but only for a massive amount of data warehousing or analysis. If your data is not that big then it could be a misfit for your company and cost you a lot. The cost associated is quite extensive as compared to some other alternative RDBMS systems available in the market.
Migration of data from Teradata to some other RDBMS systems is quite painful as the transition is not that smooth and you need to follow many steps and even if one of them fails. You need to start from the beginning almost.
Last but not least the UI is pretty outdated and needs a revamp. Though it is simple, it needs to be presented in a much better way and more advanced options need to bee presented on the front page itself.
Teradata is a mature RDBMS system that expands its functionality towards the current cloud capabilities like object storage and flexible compute scale.
Teradata Vantage allows us to create a scalable infrastructure to support our strategic initiatives. The dedicated compute power ensures reliable performance with isolated workloads and dedicated resources, optimizing workflows for faster, more efficient data transfers. The compute clusters support ETL processes and OSF’s developers and data science team with the flexibility to create self-service analytics, to spin up/down at any time, driving better performance and minimizing costs.
Community support for TensorFlow is great. There's a huge community that truly loves the platform and there are many examples of development in TensorFlow. Often, when a new good technique is published, there will be a TensorFlow implementation not long after. This makes it quick to ally the latest techniques from academia straight to production-grade systems. Tooling around TensorFlow is also good. TensorBoard has been such a useful tool, I can't imagine how hard it would be to debug a deep neural network gone wrong without TensorBoard.
We have meetings at the beginning with the technical team to explain our requirements to them and they were really putting in a lot of effort to come up with a solution which will address all our needs. They implemented the software and also trained a few of our resources on the same too. We can get in touch with them now as well whenever we run into a roadblock but it's very less now.
Keras is built on top of TensorFlow, but it is much simpler to use and more Python style friendly, so if you don't want to focus on too many details or control and not focus on some advanced features, Keras is one of the best options, but as far as if you want to dig into more, for sure TensorFlow is the right choice
Teradata is way ahead of its competitor because of its unique features of ensuring data privacy and data never gets corrupted even in worst case scenario. In most cases, the data corruption is a major issue if left unused and it leads to important data being wiped off which in ideal case should be stored for 3 years
Moving to Teradata in the Cloud-enabled a level of agility that previously didn't exist in the organization. It also enabled a level of analytic competency that was not achievable using other options on the aggressive timeline that was required. We didn't want to settle for reinventing a wheel when we had a super tuned performance capable beast readily available in Teradata. Teradata lets us focus on our business rather than spending money and effort trying to design software or database foundations features on an open source or lower performance platform.
