Google Machine Learns with ML for Sheets (ML forSpreadsheet?)

Very interesting update  ... like the idea, will look at it.   Click through for more. 

Google brings machine learning to online spreadsheets with Simple ML for Sheets

Sean Michael Kerner@TechJournalist  in Venturebeat

December 7, 2022 4:16 PM

Icon for Google Sheets  ...   Simple ML for Sheets

Check out all the on-demand sessions from the Intelligent Security Summit here.

Spreadsheets are widely used by organizations of all sizes for all kinds of basic and complex tasks.

While simple calculations and graphs have long been part of the spreadsheet experience, machine learning (ML) has not. ML is often seen as being too complex to use, while spreadsheet usage is intended to be accessible to any type of user. Google is now trying to change that paradigm for its Google Sheets online spreadsheet program.

Why adopting LC/NC tools enables an organization’s citizen developers and makes skilled developers more valuable - Low-Code/No-Code Summit

Today Google announced a beta release of the Simple ML for Sheets add-on. Google Sheets has an extensible architecture that enables users to benefit from add-ons that extend the default functionality available in the application. In this case, Google Sheets benefits from ML technology that Google first developed in the open-source TensorFlow project. With Simple ML for Sheets, users will not need to use a specific TensorFlow service, as Google has developed the service to be as easily accessible as possible.

“Everything runs completely on the user browser,” Luiz Gustavo Martins, Google AI developer advocate, told VentureBeat. “Your data doesn’t leave Google Sheets and models are saved to your Google Drive so you can use them again later.”  ... ' 

Found and Displayed from Space

Found in Space:    This image of the Cartwheel and its companion galaxies is a composite from the James Webb Space Telescope's Near-Infrared Camera (NIRCam) and Mid-Infrared Instrument (MIRI), which reveals details that are difficult to see in the individual images alone.    Credit: NASA, ESA, CSA, STScI  

By Keith Kirkpatrick,  Commissioned by CACM Staff, August 9, 2022   ... see AI tech links below.

The July 12 release of the images from NASA's James Webb Space Telescope (JWST) has captivated and excited everyone from schoolchildren to space buffs, thanks to the vivid colors and crisp captures of the distant reaches of space. The images from the telescope, which is the largest, most complex and powerful space telescope ever constructed, brought into focus thousands of galaxies, both known and unknown, as well as so-called "cosmic cliffs" of dust and gas, and even a dying star.

The telescope detects near-infrared and mid-infrared wavelengths, the light beyond the red end of the visible spectrum, which allows otherwise hidden regions of space to be captured. Infrared light can uncover and reveal new details in images, based on the object. For example, bodies of matter such as young planets that are cool and do not emit much energy or visible brightness, still radiate in the infrared. Similarly, visible light's short wavelengths often can be obscured by space dust or a dense nebula (a group of interstellar clouds), keeping their images from being captured by telescopes that only detect visible light, such as the Hubble telescope. Infrared light, with its longer wavelengths, can penetrate through dust more easily, and infrared-based telescopes can detect lower-energy objects that often form within nebulae, such as brown dwarf stars and newly forming stars. Thus, the JWST can reveal objects that previously were hidden from view.

Although the images themselves are astounding, the real value for astronomers and scientists will come from the deep analysis of the objects contained in the images. While artificial intelligence (AI) can be used to determine which data are important to be sent for processing, thereby reducing the overall amount of information that needs to be analyzed and stored, the use of deep learning provides massive benefits in the processing and analysis of the data.

Deep learning is being used to identify and classify objects from the images and can provide a significant advantage over manual classification techniques. Using a supervised approach, a training set of previously identified objects and their specific attributes and features are fed into a system to "teach" the models to yield the desired outputs. For space object classification, the training dataset includes inputs and correct outputs to identify objects such as stars, galaxies, space dust and clouds, black holes, and other elements of interest, which allow the model to learn over time. The algorithm measures its accuracy through the loss function, adjusting until the error has been sufficiently minimized to ensure confidence.

Once the model has been developed, the algorithm is ready to process, analyze, and classify new objects found by telescopes such as the JWST, saving significant amounts of time and effort over manual analysis.

"Without AI tools to perform classification, objects in astronomical images have to be inspected by professional or amateur astronomers, with a classification determined by weighting the opinions of people," says Brant Robertson, a professor of astronomy in the astrophysics department at the University of California Santa Cruz (UCSC), who is involved in the process of analyzing recently captured JWST images. "The speed of visual inspection by humans is limited by how quickly the information can be provided and by how many people can provide useful inspections of many objects. [However,] the speed of AI classification is only limited by the amount of computing available, which is no longer a considerable limitation, and the careful preparation of the datasets."

Morpheus, a deep learning framework based on TensorFlow (an end-to-end open-source platform for machine learning), will be used to perform image classification on the data captured by the JWST. Originally developed in 2019, Morpheus is a model for generating pixel-level structural classifications of astronomical data sources. It leverages deep learning to perform source detection, source segmentation, and morphological classification on a pixel-by-pixel basis, using a semantic segmentation algorithm adopted from the field of computer vision. By using this structural data about the flux of real astronomical sources during object detection, Morpheus has demonstrated resiliency to false-positive identifications of sources, according to an evaluation using data captured by the Hubble Space Telescope.

"Deep learning models like Morpheus use the full pixel information in an image to perform classification, so all the visual features of galaxies or stars are used by the model," Robertson says. "Galaxies come in three broad categories:  elliptical galaxies are ellipsoidal and relatively smooth; disk galaxies are usually flattened and have spiral structure or dark dust lanes, and irregular galaxies tend to be clumpy and amorphous. Since each of these objects have visually distinct morphologies, the model can tell them apart." Robertson says his team can "process the largest JWST surveys with Morpheus in just a few hours with the computational resources we have here at UCSC."

In a podcast conducted a few months prior to the release of the JWST images, Robertson said the telescope may be able to look for features in the atmospheres of planets that could indicate a presence of life. While Morpheus has not yet been trained to analyze this type of date, Robertson said, "We'd very much welcome collaboration from scientists interested in AI methods for evaluating JWST spectroscopic data of atmospheres."

Indeed, the study of space is a worldwide, collaborative effort, and other researchers also have developed AI platforms that can also be used to identify, evaluate, and classify objects found by space telescopes. RobERt (Robotic Exoplanet Recognition) is a deep neural network created by Ingo Waldmann and his team at the U.K.'s University College London, which used more than 85,000 simulated light curves from five classes of exoplanets to train RobERt to recognize the presence of specific molecules and gases in exoplanets' atmospheres. The platform was used to model exoplanet data from the Hubble Space Telescope, and after training, RobERt was able to identify molecules such as water, carbon dioxide, ammonia, and titanium oxide in light curves from real exoplanets with 99.7% accuracy. 

Keith Kirkpatrick is principal of 4K Research & Consulting, LLC, based in New York, NY, USA.

3D Scene Understanding

Intelligence allows us to quickly understand our world, here a good example, our visual world, based on sensors.   To establish context we can depend on.    Many new kinds of sensors now, and means of using them. 

In the Google AI Blog Technical 

3D Scene Understanding with TensorFlow 3D,   Thursday, February 11, 2021

Posted by Alireza Fathi, Research Scientist and Rui Huang, AI Resident, Google Research

The growing ubiquity of 3D sensors (e.g., Lidar, depth sensing cameras and radar) over the last few years has created a need for scene understanding technology that can process the data these devices capture. Such technology can enable machine learning (ML) systems that use these sensors, like autonomous cars and robots, to navigate and operate in the real world, and can create an improved augmented reality experience on mobile devices. The field of computer vision has recently begun making good progress in 3D scene understanding, including models for mobile 3D object detection, transparent object detection, and more, but entry to the field can be challenging due to the limited availability tools and resources that can be applied to 3D data.

In order to further improve 3D scene understanding and reduce barriers to entry for interested researchers, we are releasing TensorFlow 3D (TF 3D), a highly modular and efficient library that is designed to bring 3D deep learning capabilities into TensorFlow. TF 3D provides a set of popular operations, loss functions, data processing tools, models and metrics that enables the broader research community to develop, train and deploy state-of-the-art 3D scene understanding models.

TF 3D contains training and evaluation pipelines for state-of-the-art 3D semantic segmentation, 3D object detection and 3D instance segmentation, with support for distributed training. It also enables other potential applications like 3D object shape prediction, point cloud registration and point cloud densification. In addition, it offers a unified dataset specification and configuration for training and evaluation of the standard 3D scene understanding datasets. It currently supports the Waymo Open, ScanNet, and Rio datasets. However, users can freely convert other popular datasets, such as NuScenes and Kitti, into a similar format and use them in the pre-existing or custom created pipelines, and can leverage TF 3D for a wide variety of 3D deep learning research and applications, from quickly prototyping and trying new ideas to deploying a real-time inference system. ... " 

Google AI Describes AutoML for Time Series

 Most our careers in the big enterprise involved working with time series.   Sales, Shipments delivered, Advertising Dollars,  marketing spends ... forecast plans and predictions.   A favorite quote was 'the forecast is wrong', but how wrong?   And Why?  And what are the risks involved?  So if we could do forecasts better, more data and intelligence based?  How might we do it?     

Using AutoML for Time Series Forecasting      In the GoogleBlog.

Friday, December 4, 2020

Posted by Chen Liang and Yifeng Lu, Software Engineers, Google Research, Brain Team

Time series forecasting is an important research area for machine learning (ML), particularly where accurate forecasting is critical, including several industries such as retail, supply chain, energy, finance, etc. For example, in the consumer goods domain, improving the accuracy of demand forecasting by 10-20% can reduce inventory by 5% and increase revenue by 2-3%. Current ML-based forecasting solutions are usually built by experts and require significant manual effort, including model construction, feature engineering and hyper-parameter tuning. However, such expertise may not be broadly available, which can limit the benefits of applying ML towards time series forecasting challenges.

To address this, automated machine learning (AutoML) is an approach that makes ML more widely accessible by automating the process of creating ML models, and has recently accelerated both ML research and the application of ML to real-world problems. For example, the initial work on neural architecture search enabled breakthroughs in computer vision, such as NasNet, AmoebaNet, and EfficientNet, and in natural language processing, such as Evolved Transformer. More recently, AutoML has also been applied to tabular data.

Today we introduce a scalable end-to-end AutoML solution for time series forecasting, which meets three key criteria:

Today we introduce a scalable end-to-end AutoML solution for time series forecasting, which meets three key criteria:

Fully automated: The solution takes in data as input, and produces a servable TensorFlow model as output with no human intervention.

Generic: The solution works for most time series forecasting tasks and automatically searches for the best model configuration for each task.

High-quality: The produced models have competitive quality compared to those manually crafted for specific tasks.

We demonstrate the success of this approach through participation in the M5 forecasting competition, where this AutoML solution achieved competitive performance against hand-crafted models with moderate compute cost... .' 

Google goes for More AI Automation

More indications to the automation of the design and delivery of AI.   See the quite technical paper referenced below.The trend most likely to continue.

Google’s TF-Coder tool automates machine learning model design
Kyle Wiggers in VentureBeat    @Kyle_L_Wiggers

Researchers at Google Brain, one of Google’s AI research divisions, developed an automated tool  https://arxiv.org/pdf/2003.09040.pdf  for programming in machine learning frameworks like TensorFlow. They say it achieves better-than-human performance on some challenging development tasks, taking seconds to solve problems that take human programmers minutes to hours.

Emerging AI techniques have resulted in breakthroughs across computer vision, audio processing, natural language processing, and robotics. Playing an important role are machine learning frameworks like TensorFlow, Facebook’s PyTorch, and MXNet, which enable researchers to develop and refine new models. But while these frameworks have eased the iterating and training of AI models, they have a steep learning curve because the paradigm of computing over tensors is quite different from traditional programming. (Tensors are algebraic objects that describe relationships between sets of things related to a vector space, and they’re a convenient data format in machine learning.) Most models require various tensor manipulations for data processing or cleaning, custom loss functions, and accuracy metrics that must implemented within the constraints of a framework.  ..... " 

A Look at Transfer Learning

Good generalized look at the concept of Transfer Learnig

Everything you need to know about transfer learning in AI   in TNW

Today, artificial intelligence programs can recognize faces and objects in photos and videos, transcribe audio in real-time, detect cancer in x-ray scans years in advance, and compete with humans in some of the most complicated games.

Until a few years ago, all these challenges were either thought insurmountable, decades away, or were being solved with sub-optimal results. But advances in neural networks and deep learning, a branch of AI that has become very popular in the past few years, has helped computers solve these and many other complicated problems.

Unfortunately, when created from scratch, deep learning models require access to vast amounts of data and compute resources. This is a luxury that many can’t afford. Moreover, it takes a long time to train deep learning models to perform tasks, which is not suitable for use cases that have a short time budget.

Fortunately, transfer learning, the discipline of using the knowledge gained from one trained AI model to another, can help solve these problems.

The cost of training deep learning models
Deep learning is a subset of machine learning, the science of developing AI through training examples. The concepts and science behind deep learning and neural networks is as old as the term “artificial intelligence” itself. But until recent years, they had been largely dismissed by the AI community for being inefficient.

The availability of vast amounts of data and compute resources in the past few years have pushed neural networks into the limelight and made it possible to develop deep learning algorithms that can solve real world problems.

To train a deep learning model, you basically must feed a neural network with lots of annotated examples. These examples can be things such as labeled images of objects or mammograms scans of patients with their eventual outcomes. The neural network will carefully analyze and compare the images and develop mathematical models that represent the recurring patterns between images of a similar category.

[Read: Weird AI illustrates why algorithms still need people]

There already exists several large open-source datasets such as ImageNet, a database of more than 14 million images labeled in 22,000 categories, and MNIST, a dataset of 60,000 handwritten digits. AI engineers can use these sources to train their deep learning models.

However, training deep learning models also requires access to very strong computing resources. Developers usually use clusters of CPUs, GPUs or specialized hardware such as Google’s Tensor Processors (TPUs) to train neural networks in a time-efficient way. The costs of purchasing or renting such resources can be beyond the budget of individual developers or small organizations. Also, for many problems, there aren’t enough examples to train robust AI models.

Transfer learning makes deep learning training much less demanding
Say an AI engineer wants to create an image classifier neural network to solve a specific problem. Instead of gathering thousands and millions of images, the engineer can use one of the publicly available datasets such as ImageNet and enhance it with domain-specific photos.

But the AI engineer must still rent pay a hefty sum to rent the compute resources necessary to run those millions of images through the neural network. This is where transfer learning comes into play. Transfer learning is the process of creating new AI models by fine-tuning previously trained neural networks.  ... " 

Google releases API to train smaller, faster AI models

Interesting development, note small and faster.  I assume to make them useful for edge devices..  Also with the ability to retrain quickly.

Google releases API to train smaller, faster AI models   Kyle Wiggers  in VentureBeat

Google today released https://blog.tensorflow.org/2020/04/quantization-aware-training-with-tensorflow-model-optimization-toolkit.html Quantization Aware Training (QAT) API, which enables developers to train and deploy models with the performance benefits of quantization — the process of mapping input values from a large set to output values in a smaller set — while retaining close to their original accuracy. The goal is to support the development of smaller, faster, and more efficient machine learning models well-suited to run on off-the-shelf machines, such as those in medium- and small-business environments where computation resources are at a premium.

Often, the process of going from a higher to lower precision is noisy. That’s because quantization squeezes a small range of floating-point values into a fixed number of information buckets, leading to information loss similar to rounding errors when fractional values are represented as integers. (For example, all values in range [2.0, 2.3] might be represented in a single bucket.) Problematically, when the lossy numbers are used in several computations, the losses accumulate and need to be rescaled for the next computation. .... "

Fairness Indicators for AI Machine Learning

Brought to my attention, dealing with a potential solution for a project at hand.   Good technical and non technical coverage.  The approach is now available in Beta. Examining.

Fairness Indicators: Scalable Infrastructure for Fair ML Systems
Wednesday, December 11, 2019
Posted by Catherina Xu and Tulsee Doshi, Product Managers, Google Research

While industry and academia continue to explore the benefits of using machine learning (ML) to make better products and tackle important problems, algorithms and the datasets on which they are trained also have the ability to reflect or reinforce unfair biases. For example, consistently flagging non-toxic text comments from certain groups as “spam” or “high toxicity” in a moderation system leads to exclusion of those groups from conversation.

In 2018, we shared how Google uses AI to make products more useful, highlighting AI principles that will guide our work moving forward. The second principle, “Avoid creating or reinforcing unfair bias,” outlines our commitment to reduce unjust biases and minimize their impacts on people.

As part of this commitment, at TensorFlow World, we recently released a beta version of Fairness Indicators, a suite of tools that enable regular computation and visualization of fairness metrics for binary and multi-class classification, helping teams take a first step towards identifying unjust impacts. Fairness Indicators can be used to generate metrics for transparency reporting, such as those used for model cards, to help developers make better decisions about how to deploy models responsibly. Because fairness concerns and evaluations differ case by case, we also include in this release an interactive case study with Jigsaw’s Unintended Bias in Toxicity dataset to illustrate how Fairness Indicators can be used to detect and remediate bias in a production machine learning (ML) model, depending on the context in which it is deployed. Fairness Indicators is now available in beta for you to try for your own use cases.  ... "

TensorBoard for Free Testing of Machine Learning

Google announces a means to run machine learning experiments.  Note that the data used is pubklic and un-secured.  Includes visualizations and operation dashboards.

See:   https://tensorboard.dev/

Easily host, track, and share your ML experiments for free.
A managed TensorBoard experience that lets you upload and share your ML experiment results with anyone.

TensorBoard is TensorFlow’s visualization toolkit, enabling you to track metrics like loss and accuracy, visualize the model graph, view histograms of weights, biases, or other tensors as they change over time, and much more. It is an open source tool that is part of the TensorFlow ecosystem. 

Learn more at tensorflow.org/tensorboard.

What TensorBoard functionality is available?
In this Preview, the Scalars dashboard is currently available. Additional TensorBoard dashboards will be added over time.

I want to see the experiments I have uploaded to TensorBoard.dev. Can I get a copy of my data?
Yes. Run the command tensorboard dev export --outdir OUTPUT_PATH.

Is it free to use TensorBoard.dev?
Yes, you can view and upload TensorBoard logs for free. There are limits on how much can be stored.

Is my data public?
Any data uploaded to TensorBoard.dev will be visible to anyone with a link. Do not use it for sensitive data.

Do I need to create an account?
You need to sign in with your Google Account (or create one) to upload an experiment, so you can delete it if you want. You do not need to sign in to view an experiment.
How do I delete an experiment I've uploaded to TensorBoard.dev?
Run the command tensorboard dev delete --experiment_id EXPERIMENT_ID.

Structured Signals for Model Training

Technical but interesting point about how to add structured knowledge into otherwise non transparent networks.  Examining further.

Posted by Da-Cheng Juan (Senior Software Engineer) and Sujith Ravi (Senior Staff Research Scientist)

We are excited to introduce  Neural Structured Learning in TensorFlow, an easy-to-use framework that both novice and advanced developers can use for training neural networks with structured signals. Neural Structured Learning (NSL) can be applied to construct accurate and robust models for vision, language understanding, and prediction in general.

Neutral structured learning framework

Many machine learning tasks benefit from using structured data which contains rich relational information among the samples. For example, modeling citation networks, Knowledge Graph inference and reasoning on linguistic structure of sentences, and learning molecular fingerprints all require a model to learn from structured inputs, as opposed to just individual samples. These structures can be explicitly given (e.g., as a graph), or implicitly inferred (e.g., as an adversarial example). Leveraging structured signals during training allows developers to achieve higher model accuracy, particularly when the amount of labeled data is relatively small. Training with structured signals also leads to more robust models. These techniques have been widely used in Google for improving model performance, such as learning image semantic embedding.

Neural Structured Learning (NSL) is an open source framework for training deep neural networks with structured signals. It implements Neural Graph Learning, which enables developers to train neural networks using graphs. The graphs can come from multiple sources such as Knowledge graphs, medical records, genomic data or multimodal relations (e.g., image-text pairs). NSL also generalizes to Adversarial Learning where the structure between input examples is dynamically constructed using adversarial perturbation.  ... " 

See also:  https://www.datanami.com/2019/09/04/google-adds-structured-signals-to-model-training/

See also:  https://venturebeat.com/2019/09/03/google-launches-tensorflow-machine-learning-framework-for-graphical-data/ 

Accelerating AI with Open Source, and More

Update on MLIR, which we had looked at.   See who has joined the consortium.  Architecture always being a key element to doing anything well.   And to do things efficiently it makes lots of sense to share the work.   I would further add there should be better shared ways to manage varying data  'infrastructures' by problem domains, in both the semantics of the data and its metadata.   Lets make that happen too.

MLIR: accelerating AI with open-source infrastructure

Chris Lattner, Distinguished Engineer, TensorFlow

Tim Davis,  Product Manager, TensorFlow

Machine learning now runs on everything from cloud infrastructure containing GPUs and TPUs, to mobile phones, to even the smallest hardware like microcontrollers that power smart devices. The combination of advancements in hardware and open-source software frameworks like TensorFlow is making all of the incredible AI applications we’re seeing today possible--whether it’s predicting extreme weather, helping people with speech impairments communicate better, or assisting farmers to detect plant diseases. 

But with all this progress happening so quickly, the industry is struggling to keep up with making different machine learning software frameworks work with a diverse and growing set of hardware. The machine learning ecosystem is dependent on many different technologies with varying levels of complexity that often don't work well together. The burden of managing this complexity falls on researchers, enterprises and developers. By slowing the pace at which new machine learning-driven products can go from research to reality, this complexity ultimately affects our ability to solve challenging, real-world problems. 

Earlier this year we announced MLIR, open source machine learning compiler infrastructure that addresses the complexity caused by growing software and hardware fragmentation and makes it easier to build AI applications. It offers new infrastructure and a design philosophy that enables machine learning models to be consistently represented and executed on any type of hardware. And today we’re announcing that we’re contributing MLIR to the nonprofit LLVM Foundation. This will enable even faster adoption of MLIR by the industry as a whole.   .... " 

Speeding Up Learning Inference by 2X

New methods, technical:

New Technique Speeds Up Deep-Learning Inference on TensorFlow by 2x
by  Anthony Alford  in InfoQ

Researchers at North Carolina State University recently presented a paper at the International Conference on Supercomputing (ICS) on their new technique, "deep reuse" (DR), that can speed up inference time for deep-learning neural networks running on TensorFlow by up to 2x, with almost no loss of accuracy.

Dr. Xipeng Shen, along with graduate student Lin Ning, authored the paper describing the technique, which requires no special hardware or changes to the deep-learning model. By taking advantage of similarities in the data values that are input into a neural network layer, DR eliminates redundant computation during inference, reducing the total time taken. Reducing computation also reduces power consumption, a key feature for mobile or embedded applications. In experiments running several common computer-vision deep-learning models on GPUs, including CifarNet, AlexNet, and VGG-19, DR achieved from 1.75X to 2.02X speedup, with an increase in error of 0.0005. In some cases, DR actually improved accuracy slightly. In similar experiments on a mobile phone, DR "achieves an average of 2.12x speedup for CifarNet and 2.55X for AlexNet."  .... "

Free eBook on TensorFlow in the Enterprise

Should you use TensorFlow in your enterprise? via O'Reilly
Find out with this free ebook

TensorFlow World is where you stay ahead on the latest in TensorFlow & machine learning. Join us October 28-31 in Santa Clara.

The question is no longer whether your enterprise will use deep learning (you will), but how involved your company will be with the technology.

If your company is adopting deep learning, this short ebook, Considering TensorFlow for the Enterprise, will help you navigate the initial decisions you must make—from choosing a deep learning framework to integrating deep learning with the other data analysis systems already in place—to ensure you're building a system capable of handling your specific business needs.

And it’s yours, free. ...

Google What-IF Tool for Code Free ML Visualization

Like the idea of visual tools that map with specific process, resource needs and output results.  Leads to better understandable and resilient results.

Google's What-If Tool And The Future Of Explainable AI
Kalev Leetaru Contributor in Forbes
AI & Big Data

(Excerpt)

" ..... As deep learning has matured sufficiently to find widespread adoption in industry and as developers require increasingly greater understanding of their creations in order to pioneer new advances, the AI community has begun investing heavily in explainable AI as a way to render their black boxes transparent.

Google has been an early leader in emphasizing interpretability and how practitioners can build more understandable, representative and resilient AI solutions. Last year the company unveiled its What-If Tool, which offers a range of interactive visualizations and guided explorations of a TensorFlow model, allowing developers to explore how their model interpreted its training data and how subtle changes to a given input would change its classification, yielding insights into the model’s robustness. .... " 

Google's Description:

The What-If Tool: Code-Free Probing of Machine Learning Models
Tuesday, September 11, 2018
Posted by James Wexler, Software Engineer, Google AI

Building effective machine learning (ML) systems means asking a lot of questions. It's not enough to train a model and walk away. Instead, good practitioners act as detectives, probing to understand their model better: How would changes to a datapoint affect my model’s prediction? Does it perform differently for various groups–for example, historically marginalized people? How diverse is the dataset I am testing my model on?

Answering these kinds of questions isn’t easy. Probing “what if” scenarios often means writing custom, one-off code to analyze a specific model. Not only is this process inefficient, it makes it hard for non-programmers to participate in the process of shaping and improving ML models. One focus of the Google AI PAIR initiative is making it easier for a broad set of people to examine, evaluate, and debug ML systems.


Today, we are launching the What-If Tool, a new feature of the open-source TensorBoard web application, which let users analyze an ML model without writing code. Given pointers to a TensorFlow model and a dataset, the What-If Tool offers an interactive visual interface for exploring model results.  .... " 

Tensorflow 2.0

Introduction to the announcement, not too technical.

Now TensorFlow 2.0 and its high-level APIs (via TF Dev Summit '19)
With TensorFlow 2.0, we are consolidating our APIs and integrating Keras across the TensorFlow ecosystem. In this talk, we give an overview of what to expect with TensorFlow High Level APIs in 2.0.  ... 

See the revamped dev site  → https://www.tensorflow.org/
Watch all TensorFlow Dev Summit '19 sessions → http://bit.ly/TFDS19Sessions
Subscribe to the TensorFlow YouTube channel → https://bit.ly/TensorFlow1

TensorFlow

Was asked this question recently.  Here a quick, non technical answer.  But does also include code, which is by its nature technical.

What is Tensorflow?

ODSC    https://opendatascience.com/  

It would be a challenge nowadays to find a machine learning engineer who has heard nothing about TensorFlow. Initially created by Google Brain team for some internal purposes, such as spam filtering on Gmail, it was open-sourced in 2015 and became the most popular deep learning framework in the next few years.

Tensorflow is often used for solving deep learning problems and for training and evaluating processes up to the model deployment. Apart from machine learning purposes, TensorFlow can be also used for building simulations, based on partial derivative equations. That’s why it is considered to be an all-purpose tool for machine learning engineers.  ... " 

Google has an AI Cloud Platform. Lets link it with BPM

Quite some detail for making AI applications work with the cloud in this new production factory for AI in the Cloud.  I like the idea of standardizing such learning projects and installed solutions. I would also like to see this kind of work linked with business process models like BPM.

AI Platform

Create your AI applications once, then run them easily on both GCP and on-premises.

Take your machine learning projects to production

AI Platform makes it easy for machine learning developers, data scientists, and data engineers to take their ML projects from ideation to production and deployment, quickly and cost-effectively. From data engineering to “no lock-in” flexibility, AI Platform’s integrated tool chain helps you build and run your own machine learning applications.

AI Platform supports Kubeflow, Google’s open-source platform, which lets you build portable ML pipelines that you can run on-premises or on Google Cloud without significant code changes. And you’ll have access to cutting-edge Google AI technology like TensorFlow, TPUs, and TFX tools as you deploy your AI applications to production.  ... "

A testimonial they provide:

" ... In retail, it’s important to provide customers with easy access to alternative products or recommended add-ons. We train our own machine learning models with TensorFlow on Google Cloud ML, and we automate the periodic retraining of these models with Kubeflow Pipelines. Together with AI Hub, useful for sharing models between data scientists, we can now iterate faster on our models, and automatically deploy them to staging and production. ... '    Lucas Ngoo, co-founder, CTO, Carousell

See also: https://techcrunch.com/2019/04/10/google-expands-its-ai-services/

Google Celebrates Bach Birthday with AI

Nicely done, with a basic description about how current AI systems work.   And some creative and playful Bach arrangements as well!   All guided by a learning database of 306 genuine Bach compositions.  All at the link.

Google writes:

Today we celebrate world renowned German composer and musician Johann Sebastian Bach with our first ever AI-powered Doodle! Made in partnership with the Google Magenta and Google PAIR teams, the Doodle is an interactive experience encouraging players to compose a two measure melody of their choice. With the press of a button, the Doodle then uses machine learning to harmonize the custom melody into Bach’s signature music style (or a Bach 80's rock style hybrid if you happen to find a very special easter egg in the Doodle...:)). 

The first step in developing the Doodle? Creating a machine learning model to power it. Machine learning is the process of teaching a computer to come up with its own answers by showing it a lot of examples, instead of giving it a set of rules to follow as is done in traditional computer programming. The model used in today's Doodle was developed by Magenta Team AI Resident Anna Huang, who developed Coconet: a versatile model that can be used in a wide range of musical tasks—such as harmonizing melodies or composing from scratch (check out more of these technical details in today’s Magenta blog post).

Specifically, Coconet was trained on 306 of Bach’s chorale harmonizations. His chorales always have four voices, each carrying their own melodic line, while creating a rich harmonic progression when played together. This concise structure made them good training data for a machine learning model. 

Next came our partners at PAIR who used TensorFlow.js to allow machine learning to happen entirely within the web browser (versus it running utilizing tons of servers, as machine learning traditionally does). For cases where someone’s computer or device might not be fast enough to run the Doodle using TensorFlow.js, the Doodle is also served with Google’s new Tensor Processing Units (TPUs), a way of quickly handling machine learning tasks in data centers— yet another Doodle first!

These components, combined with art and engineering from the Doodle team, helped create what you see today.... "

Beyond Musical Improvisation

Have always mused that improvisation should exist beyond music.   Business process extension that lead to better results?  Creative extension of the basic?  Note AI connection.   Real time aspects for business?  Here an example in the musical realm that lets you think about it.  Technical explanation and online demo at the link:

Piano Genie: An Intelligent Musical Interface

We introduce Piano Genie, an intelligent controller that maps 8-button input to a full 88-key piano in real time:

Piano Genie is in some ways reminiscent of video games such as Rock Band and Guitar Hero that are accessible to novice musicians, with the crucial difference that users can freely improvise on Piano Genie rather than re-enacting songs from a fixed repertoire. You can try it out yourself via our interactive web demo! .... " 

Analyze an ML Model without More Coding

Once a system has been taught it acts as an experimental model of its learning.  When we built such neural models you had to write code or use data manipulation tools to test against new data sets.  Classic method for all analytic models, not only ML.  This new tool should make it easier.

Google's new What-If Tool "allows users to analyze a machine learning model without the need for writing any further code. Given pointers to a TensorFlow model and a dataset, the What-If Tool offers an interactive visual interface for exploring model results."

What If...
you could inspect a machine learning model, with no coding required?
Building effective machine learning systems means asking a lot of questions. It's not enough to train a model and walk away. Instead, good practitioners act as detectives, probing to understand their model better.

But answering these kinds of questions isn't easy. Probing "what if" scenarios often means writing custom, one-off code to analyze a specific model. Not only is this process inefficient, it makes it hard for non-programmers to participate in the process of shaping and improving machine learning models. For us, making it easier for a broad set of people to examine, evaluate, and debug machine learning systems is a key concern.

That's why we built the What-If Tool. Built into the open-source TensorBoard web application - a standard part of the TensorFlow platform - the tool allows users to analyze an machine learning model without the need for writing any further code. Given pointers to a TensorFlow model and a dataset, the What-If Tool offers an interactive visual interface for exploring model results.... "