Vector Search: The Tech Powering Billions of Search Results for Google Users
4543
Of your peers have already read this article.
1:30 Minutes
The most insightful time you'll spend today!
Recently, Google Cloud partner Groovenauts, Inc. published a live demo of MatchIt Fast. As the demo shows, you can find images and text similar to a selected sample from a collection of millions in a matter of milliseconds:
Give it a try — and either select a preset image or upload one of your own. Once you make your choice, you will get the top 25 similar images from two million images on Wikimedia images in an instant, as you can see in the video above. No caching involved.
The demo also lets you perform the similarity search with news articles. Just copy and paste some paragraphs from any news article, and get similar articles from 2.7 million articles on the GDELT project within a second.
Vector Search: the technology behind Google Search, YouTube, Play, and more
How can it find matches that fast? The trick is that the MatchIt Fast demo uses the vector similarity search (or nearest neighbor search or simply vector search) capabilities of the Vertex AI Matching Engine, which shares the same backend as Google Image Search, YouTube, Google Play, and more, for billions of recommendations and information retrievals for Google users worldwide. The technology is one of the most important components of Google’s core services, and not just for Google: it is becoming a vital component of many popular web services that rely on content search and information retrieval accelerated by the power of deep neural networks.
So what’s the difference between traditional keyword-based search and vector similarity search? For many years, relational databases and full-text search engines have been the foundation of information retrieval in modern IT systems. For example, you would add tags or category keywords such as “movie”, “music”, or “actor” to each piece of content (image or text) or each entity (a product, user, IoT device, or anything really). You’d then add those records to a database, so you could perform searches with those tags or keywords.
In contrast, vector search uses vectors (where each vector is a list of numbers) for representing and searching content. The combination of the numbers defines similarity to specific topics. For example, if an image (or any content) includes 10% of “movie”, 2% of “music”, and 30% of “actor”-related content, then you could define a vector [0.1, 0.02, 0.3] to represent it. (Note: this is an overly simplified explanation of the concept; the actual vectors have much more complex vector spaces). You can find similar content by comparing the distances and similarities between vectors. This is how Google services find valuable content for a wide variety of users worldwide in milliseconds.
With keyword search, you can only specify a binary choice as an attribute of each piece of content; it’s either about a movie or not, either music or not, and so on. Also, you cannot express the actual “meaning” of the content to search. If you specify a keyword “films”, for example, you would not see any content related to “movies” unless there was a synonyms dictionary that explicitly linked these two terms in the database or search engine.
Vector search provides a much more refined way to find content, with subtle nuances and meanings. Vectors can represent a subset of content that contains “much about actors, some about movies, and a little about music”. Vectors can represent the meaning of content where “films”, “movies”, and “cinema” are all collected together. Also, vectors have the flexibility to represent categories previously unknown to or undefined by service providers. For example, emerging categories of content primarily attractive to kids, such as ASMR or slime, are really hard for adults or marketing professionals to predict beforehand, and going back through vast databases to manually update content with these new labels would be all but impossible to do quickly. But vectors can capture and represent never-before-seen categories instantly.
Vector search changes business
Vector search is not only applicable to image and text content. It can also be used for information retrieval for anything you have in your business when you can define a vector to represent each thing. Here are a few examples:
- Finding similar users: If you define a vector to represent each user in your business by combining the user’s activities, past purchase history, and other user attributes, then you can find all users similar to a specified user. You can then see, for example, users who are purchasing similar products, users that are likely bots, or users who are potential premium customers and who should be targeted with digital marketing.
- Finding similar products or items: With a vector generated from product features such as description, price, sales location, and so on, you can find similar products to answer any number of questions; for example, “What other products do we have that are similar to this one and may work for the same use case?” or “What products sold in the last 24 hours in this area?” (based on time and proximity)
- Finding defective IoT devices: With a vector that captures the features of defective devices from their signals, vector search enables you to instantly find potentially defective devices for proactive maintenance.
- Finding ads: Well-defined vectors let you find the most relevant or appropriate ads for viewers in milliseconds at high throughput.
- Finding security threats: You can identify security threats by vectorizing the signatures of computer virus binaries or malicious attack behaviors against web services or network equipment.
- …and many more: Thousands of different applications of vector search in all industries will likely emerge in the next few years, making the technology as important as relational databases.
OK, vector search sounds cool. But what are the major challenges to applying the technology to real business use cases? Actually there are two:
- Creating vectors that are meaningful for business use cases
- Building a fast and scalable vector search service
Embeddings: meaningful vectors for business use cases
The first challenge is creating vectors for representing various entities that are meaningful and useful for business use cases. This is where deep learning technology can really shine. In the case of the MatchIt Fast demo, the application simply uses a pre-trained MobileNet v2 model for extracting vectors from images, and the Universal Sentence Encoder (USE) for text. By applying such models to raw data, you can extract “embeddings” – vectors that map each row of data in a space of their “meanings”. MobileNet puts images that have similar patterns and textures closer to one another in the embedding space, and USE puts texts that have similar topics closer.
For example, a carefully designed and trained machine learning model could map movies into an embedding space like the following:
With the embedding space shown here, users could find recommended movies based on the two dimensions: is the movie for children or adults, and is it a blockbuster or arthouse movie? This is a very simple example, of course, but with an embedding space like this that fits your business requirements, you can deliver a better user experience on recommendation and information retrieval services with insights extracted from the model.
For more about creating embeddings, the Machine Learning Crash Course on Recommendation Systems is a great way to get started. We will also discuss how to extract better embeddings from business data later in this post.
Building a fast and scalable vector search service
Suppose that you have successfully extracted useful vectors (embeddings) from your business data. Now the only thing you have to do is search for similar vectors. That sounds simple, but in practice it is not. Let’s see how the vector search works when you implement it with BigQuery in a naive way:https://www.youtube.com/embed/wHNJspvxj2w?enablejsapi=1&
It takes about 20 seconds to find similar items (fish images in this case) from a pool of one million items. That level of performance is not so impressive, especially when compared to the MatchIt Fast demo. BigQuery is one of the fastest data warehouse services in the industry, so why does the vector search take so long?
This illustrates the second challenge: building a fast and scalable vector search engine isn’t an easy task. The most widely used metrics for calculating the similarity between vectors are L2 distance (Euclidean distance), cosine similarity, and inner product (dot product).
But all require calculations proportional to the number of vectors multiplied by the number of dimensions if you implement them in a naive way. For example, if you compare a vector with 1024 elements to 1M vectors, the number of calculations will be proportional to 1024 x 1M = 1.02B. This is the computation required to look through all the entities for a single search, and the reason why the BigQuery demo above takes so long.
Instead of comparing vectors one by one, you could use the approximate nearest neighbor (ANN) approach to improve search times. Many ANN algorithms use vector quantization (VQ), in which you split the vector space into multiple groups, define “codewords” to represent each group, and search only for those codewords. This VQ technique dramatically enhances query speeds and is the essential part of many ANN algorithms, just like indexing is the essential part of relational databases and full-text search engines.
As you may be able to conclude from the diagram above, as the number of groups in the space increases the speed of the search decreases and the accuracy increases. Managing this trade-off — getting higher accuracy at shorter latency — has been a key challenge with ANN algorithms.
Last year, Google Research announced ScaNN, a new solution that provides state-of-the-art results for this challenge. With ScaNN, they introduced a new VQ algorithm called anisotropic vector quantization:
Anisotropic vector quantization uses a new loss function to train a model for VQ for an optimal grouping to capture farther data points (i.e. higher inner product) in a single group. With this idea, the new algorithm gives you higher accuracy at lower latency, as you can see in the benchmark result below (the violet line):
This is the magic ingredient in the user experience you feel when you are using Google Image Search, YouTube, Google Play, and many other services that rely on recommendations and search. In short, Google’s ANN technology enables users to find valuable information in milliseconds, in the vast sea of web content.
How to use Vertex AI Matching Engine
Now you can use the same search technology that powers Google services with your own business data. Vertex AI Matching Engine is the product that shares the same ScaNN based backend with Google services for fast and scalable vector search, and recently it became GA and ready for production use. In addition to ScaNN, Matching Engine gives you additional features as a commercial product, including:
- Scalability and availability: The open source version of ScaNN is a good choice for evaluation purposes, but as with most new and advanced technologies, you can expect challenges when putting it into production on your own. For example, how do you operate it on multiple nodes with high scalability, availability, and maintainability? Matching Engine uses Google’s production backend for ScaNN, which provides auto-scaling and auto-failover with a large worker pool. It is capable of handling tens of thousands of requests per second, and returns search results in less than 10 ms for the 90th percentile with a recall rate of 95 – 98%.
- Fully managed: You don’t have to worry about building and maintaining the search service. Just create or update an index with your vectors, and you will have a production-ready ANN service deployed. No need to think about rebuilding and optimizing indexes, or other maintenance tasks.
- Filtering: Matching Engine provides filtering functionality that enables you to filter search results based on tags you specify on each vector. For example, you can assign “country” and “stocked” tags to each fashion item vector, and specify filters like “(US OR Canada) AND stocked” or “not Japan AND stocked” on your searches.
Let’s see how to use Matching Engine with code examples from the MatchIt Fast demo.
Generating embeddings
Before starting the search, you need to generate embeddings for each item like this one:
{"Id":"b5c65fea9b0b8a57bfa574ea","Embedding": [0.16329009830951691,0.92436742782592773,0.00095699273515492678,0.011479727923870087,0.0089491046965122223,0.019959751516580582,0.031516745686531067,0.0066015380434691906,0.46404418349266052,...
This is an embedding with 1280 dimensions for a single image, generated with a MobileNet v2 model. The MatchIt Fast demo generates embeddings for two million images with the following code:
class Vectorizer:def __init__(self):self._model = tf.keras.Sequential([hub.KerasLayer("https://tfhub.dev/google/imagenet/mobilenet_v2_100_224/feature_vector/5", trainable=False)])self._model.build([None, 224, 224, 3]) # Batch input shape.def vectorize(self, jpeg_file):...snip...embedding = self._model.predict({"inputs": input_tensor})[0].tolist()return embedding
After you generate the embeddings, you store them in a Google Cloud Storage bucket.
Configuring an index
Then, define a JSON file for the index configuration:
{"contentsDeltaUri": "gs://match-it-fast-us-central1/wikimedia_images/index-1","config": {"dimensions": 1280,"approximateNeighborsCount": 150,"distanceMeasureType": "SQUARED_L2_DISTANCE","algorithm_config": {"treeAhConfig": {"leafNodeEmbeddingCount": 1000,"leafNodesToSearchPercent": 5}}}}
You can find a detailed description for each field in the documentation, but here are some important fields:
contentsDeltaUri: the place where you have stored the embeddingsdimensions: how many dimensions in the embeddingsapproximateNeighborsCount: the default number of neighbors to find via approximate searchdistanceMeasureType: how the similarity between embeddings should be measured, either L1, L2, cosine or dot product (this page explains which one to choose for different embeddings)
To create an index on the Matching Engine, run the following gcloud command where the metadata-file option takes the JSON file name defined above.
gcloud --project=gn-match-it-fast beta ai indexes create \--display-name=wikimedia-images \--description="Wikimedia Image Demo" \--metadata-file=metadata/wikimedia_images_index_metadata.json \--region=us-central1
Run the search
Now the Matching Engine is ready to run. The demo processes each search request in the following order:
- First, the web UI takes an image (the one chosen or uploaded by the user) and encodes it into an embedding using the TensorFlow.js MobileNet v2 model running inside the browser. Note: this “client-side encoding” is an interesting option for reducing network traffic when you can run the encoding at the client. In many other cases, you would encode contents to embeddings with a server-side prediction service such as Vertex AI Prediction, or just retrieve pre-generated embeddings from a repository like Vertex AI Feature Store.
- The App Engine frontend receives the embedding and submits a query to the Matching Engine. Note that you can also use any other compute services in Google Cloud for submitting queries to Matching Engine, such as Cloud Run, Compute Engine, or Kubernetes Engine, or whatever is most suitable for your applications.
- Matching Engine executes its search. The connection between App Engine and Matching Engine is provided via a VPC private network for optimal latency.
- Matching Engine returns the IDs of similar vectors in its index.
Step 3 is implemented with the following code:
class MatchingQueryClient:...snip...def query_embedding(self, embedding, num_neighbors=30):request = match_service_pb2.MatchRequest()request.deployed_index_id = self._deployed_index_idfor v in embedding:request.float_val.append(v)request.num_neighbors = num_neighborsresponse = self._stub.Match(request)return response
The request to the Matching Engine is sent via gRPC as you can see in the code above. After it gets the request object, it specifies the index id, appends elements of the embedding, specifies the number of neighbors (similar embeddings) to retrieve, and calls the Match function to send the request. The response is received within milliseconds.
Next steps: Making changes for various use cases and better search quality
As we noted earlier, the major challenges in applying vector search on production use cases are:
- Creating vectors that are meaningful for business use cases
- Building a fast and scalable vector search service
From the example above, you can see that Vertex AI Matching Engine solves the second challenge. What about the first one? Matching Engine is a vector search service; it doesn’t include the creating vectors part.
The MatchIt Fast demo uses a simple way of extracting embeddings from images and contents; specifically it uses an existing pre-trained model (either MobileNet v2 or Universal Sentence Encoder). While those are easy to get started with, you may want to explore other options to generate embeddings for other use cases and better search quality, based on your business and user experience requirements.
For example, how do you generate embeddings for product recommendations? The Recommendation Systems section of the Machine Learning Crash Course is a great resource for learning how to use collaborative filtering and DNN models (the two-tower model) to generate embeddings for recommendation. Also, TensorFlow Recommenders provides useful guides and tutorials for the topic, especially on the two-tower model and advanced topics. For integration with Matching Engine, you may also want to check out the Train embeddings by using the two-tower built-in algorithm page.
Another interesting solution is the Swivel model. Swivel is a method for generating item embeddings from an item co-occurrence matrix. For structured data, such as purchase orders, the co-occurrence matrix of items can be computed by counting the number of purchase orders that contain both product A and product B, for all products you want to generate embeddings for. To learn more, take a look at this tutorial on how to use the model with Matching Engine.
If you are looking for more ways to achieve better search quality, consider metric learning, which enables you to train a model for discrimination between entities in the embedding space, not only classification:
Popular pre-trained models such as the MobileNet v2 can classify each object in an image, but they are not explicitly trained to discriminate the objects from each other with a defined distance metric. With metric learning, you can expect better search quality by designing the embedding space optimized for various business use cases. TensorFlow Similarity could be an option for integrating metric learning with Matching Engine.
Interested? Today, we’re just beginning the migration from traditional search technology to new vector search. Over the next 5 to 10 years, many more best practices and tools will be developed in the industry and community. These tools and best practices will help answer many questions, like… How do you design your own embedding space for a specific business use case? How do you measure search quality? How do you debug and troubleshoot the vector search? How do you build a hybrid setup with existing search engines for meeting sophisticated requirements? There are many new challenges and opportunities ahead for introducing the technology to production. Now’s the time to get started delivering better user experiences and seizing new business opportunities with Matching Engine powered by vector search.
Acknowledgements
We would like to thank Anand Iyer, Phillip Sun, and Jeremy Wortz for their invaluable feedback to this post.
Discover Latest Resources on Google Cloud’s Datasets Solution
6325
Of your peers have already read this article.
3:00 Minutes
The most insightful time you'll spend today!
Editor’s note: With Google Cloud’s datasets solution, you can access an ever-expanding resource of the newest datasets to support and empower your analyses and ML models, as well as frequently updated best practices on how to get the most out of any of our datasets. We will be regularly updating this blog with new datasets and announcements, so be sure to bookmark this link and check back often.
August 2021
New dataset: Google Cloud Release Notes
- Looking for a new way to access Google Cloud Release Notes besides the docs, the XML feed, and Cloud Console? Check out the Google Cloud Release Notes dataset. With up-to-date release notes for all generally available Google Cloud products, this dataset allows you to use BigQuery to programmatically analyze release notes across all products, exploring security bulletin updates, fixes, changes, and the latest feature releases.
- Access the dataset
July 2021
Best practice: Use Google Trends data for common business needs
- The Google Trends dataset represents the first time we’re adding Google-owned Search data into Datasets for Google Cloud. The Trends data allows users to measure interest in a particular topic or search term across Google Search, from around the United States, down to the city-level. You can learn more about the dataset here, and check out the Looker dashboard here! These tables are super valuable in their own right, but when you blend them with other actionable data you can unlock whole new areas of opportunity for your team. To learn how to make informed decisions with Google Trends data, keep reading.
- Access the dataset
New dataset: COVID-19 Vaccination Search Insights
- With COVID-19 vaccinations being a topic of interest around the United States, this dataset shows aggregated, anonymized trends in searches related to COVID-19 vaccination and is intended to help public health officials design, target, and evaluate public education campaigns. Check out this interactive dashboard to explore searches for COVID-19 vaccination topics by region.
- Access the dataset
June 2021
New dataset: Google Diversity Annual Report 2021
- Since 2014, Google has disclosed data on the diversity of its workforce in an effort to bring candid transparency to the challenges technology companies like Google face in recruitment and retention of underrepresented communities. In an effort to make this data more accessible and useful, we’ve loaded it into BigQuery for the first time ever. To view Google’s Diversity Annual Report and learn more, check it out.
- Access the dataset
New dataset: Google Trends Top 25 Search terms
- The most popular and surging Google Search terms are now available in BigQuery as a public dataset. View the Top 25 and Top 25 rising queries from Google Trends from the past 30-days, including 5 years of historical data across the 210 Designated Market Areas (DMAs) in the US. Keep reading.
- Access the dataset
New dataset: COVID-19 Vaccination Access
- With metrics quantifying travel times to COVID-19 vaccination sites, this dataset is intended to help Public Health officials, researchers, and Healthcare Providers to identify areas with insufficient access, deploy interventions, and research these issues. Check out how this data is being used in a number of new tools.
- Access the dataset
Best practice: Leveraging BigQuery Public Boundaries datasets for geospatial analytics
- Geospatial data is a critical component for a comprehensive analytics strategy. Whether you are trying to visualize data using geospatial parameters or do deeper analysis or modeling on customer distribution or proximity, most organizations have some type of geospatial data they would like to use – whether it be customer zipcodes, store locations, or shipping addresses. However, converting geographic data into the correct format for analysis and aggregation at different levels can be difficult. In this post, we’ll walk through some examples of how you can leverage the Google Cloud platform alongside Google Cloud Public Datasets to perform robust analytics on geographic data. Keep reading.
- Access the dataset
Get the metadata and try BigQuery sandbox
When you’ve learned about many of our datasets and pre-built solutions from across Google, you may be ready to start querying them. Check out the full dataset directory and read all the metadata at g.co/cloud/marketplace-datasets, then dig into the data with our free-to-use BigQuery sandbox account, or $300 in credits with our Google Cloud free trial.
What Are India’s Biggest Companies Doing on Google Cloud?
9276
Of your peers have already read this article.
6:30 Minutes
The most insightful time you'll spend today!
In the last year, there’s been an upward trend in cloud adoption in India. In fact, NASSCOM finds that cloud spending in India is estimated to grow at 30% per annum to cross the US$7 billion mark by 2022.
At Google, in our conversations with customers, discussions have evolved beyond cost savings and efficiencies. While those are still very relevant reasons for adopting cloud technologies, Indian enterprises are looking to Google Cloud to help them drive digital transformation, identify new revenue generating business models, reach previously untapped consumer markets, and build customer loyalty through greater insight and personalization.
Here are some companies and their stories.
Tata Steel: Mining data and maximizing its power
Tata Steel is a great example of an established enterprise from a traditional industry that is modernizing and embracing cloud computing. With an ambition to be a leader in manufacturing in India and a digital-first organization by 2022, Tata Steel believes smart analytics is key to enhancing operational efficiency and gaining business advantage.
To organize data from siloed systems across the organization and make it easily accessible to all employees, Tata Steel is using Cloud Search and plans to scale it to more than one million documents and 28 disparate enterprise content sources including enterprise resource planning (ERP) and SharePoint. In fact, Tata Steel is one of the first Indian enterprises to harness the power of Cloud Search to meet some of the most aggressive ingestion demands, with indexing durations reduced from weeks to seconds.
They are also leveraging Google Cloud Platform (GCP) services like Google Cloud Storage and BigQuery to build their data lake and enterprise data warehouse so they can take advantage of advanced analytics and machine learning. Managed services such as AI Platform further enable Tata Steel to manage end-to-end AI/ML workflows within the GCP console. This complements their existing on-premise reporting and analytics tools, and brings data management to the forefront of everything they do—from forecasting market demand to predictive equipment maintenance.
“Digital is not just a goal, it’s become a way of life. We are digitizing everything from the deployment of factory vehicles to improving material throughput to marketing and sales. As a result, we have petabytes of structured and unstructured data that is not only waiting to be mined, but that we can generate intelligence from to create opportunities across our multiple lines of business using GCP,” said Sarajit Jha, Chief Business Transformation & Digital Solutions at Tata Steel.
Helping L&T Financial Services reach customers in rural communities
In rural communities, quick access to financial services can make a tremendous difference to livelihoods. L&T Financial Services provides farm-equipment finance, micro loans and two-wheeler finance to consumers across rural India backed by a strong digital and analytics platform. Their digital-loan approval app, which runs on GCP, makes it significantly faster and easier for people to apply for financial assistance to purchase important things such as farming equipment and two-wheelers. It also helps rural women entrepreneurs get quicker access to funds for their businesses through micro loans.
L&T Financial found G Suite to be a far better collaborative tool to help staff work together efficiently. Employees can interact with each other in real time using Hangouts Meet, and the task of information sharing is more seamless and secure through Drive. BigQuery also helps L&T Financial Services generate behavior scorecards to track credit quality of its micro-loan customers.
“Cloud is the technology that enables us to achieve scale and reach. Today there are countless data points available about rural consumers which enable us to personalize our products to serve them better. With access to faster compute power, we can also on-board consumers more efficiently. Our rural businesses have clocked a disbursement CAGR of 60% over the past three years.” said Sunil Prabhune, Chief Executive-Rural Finance, and Group Head-Digital, IT and Analytics, L&T Financial Services.
Creating conversational connections for Digitate’s customers
Digitate, a venture of TCS (Tata Consultancy Services), has integrated Dialogflow into its flagship brand ignio, an award-winning artificial intelligence platform for driving IT operations, workload operations and ERP operations for diverse enterprises. This integration is the next step in ignio’s product development journey, and will enable users to chat or talk with ignio to detect issues, triage problems, resolve them and even predict system behavior.
“ignio combines its unique self-healing AIOps capabilities for enterprise IT and business operations with Dialogflow’s AI/ML-based, easy to use, natural and rich conversational capabilities to create an unparalleled, intuitive and feature-rich experience for our customers,” says Akhilesh Tripathi, Head of Digitate.
Indian enterprises going G Suite
The base of Indian enterprises that are making the switch to G Suite to streamline their productivity and collaboration also continues to grow. Sharechat, BookMyShow, Hero MotorCorp, DB Corp and Royal Enfield are now able to move faster within their organizations, using intelligent, cloud-based apps to transform the way they work.
A hybrid and multi-cloud future in India
IDC predicts that by 2023, 55% of India 500 organizations will have a multi-cloud management strategy that includes integrated tools across public and private clouds. (IDC FutureScape: Worldwide Cloud 2019 Predictions — India Implications (# AP43922319). We look forward to sharing more success stories of Indian enterprises that have taken the next step in their digital transformation journey.
Enhanced AI-based Photo Editing: Let’s Enhance Fuels The Next Wave of Innovation with Google Cloud and NVIDIA
2710
Of your peers have already read this article.
3:30 Minutes
The most insightful time you'll spend today!
There’s an explosion in the number of digital images generated and used for both personal and business needs. On e-commerce platforms and online marketplaces for example, product images and visuals heavily influence the consumer’s perception, decision making and ultimately conversion rates. In addition, there’s been a rapid shift towards user-generated visual content for ecommerce — think seller-generated product imagery, host-generated rental property photos and influencer-generated social media content. The challenge? These user-generated images are often captured using mobile cameras and vary greatly in terms of their size, quality, compression ratios and resolution, making it difficult for companies to provide consistent high-quality product images on their platforms.
That’s exactly the problem that Let’s Enhance, a computer vision startup with teams across the US and Ukraine, set out to solve with AI. The Let’s Enhance platform improves the quality of any user-generated photo automatically using AI-based features to enhance images through automatic upscaling, pixelation and blur fixes, color and low-light correction, and removing compression artifacts — all with a single click and no professional equipment or photo-editing chops.
“Let’s Enhance.io is designed to be a simple platform that brings AI-powered visual technologies to everyone — from marketers and entrepreneurs to photographers and designers,” said Sofi Shvets, CEO and Co-founder of Let’s Enhance.
To date, Let’s Enhance has processed more than 100M photos for millions of customers worldwide for use-cases ranging from digital art galleries, real estate agencies, digital printing, ecommerce and online marketplaces. With the introduction of Claid.ai, their new API to automatically enhance and optimize user-generated content at scale for digital marketplaces, they needed to process millions of images every month and manage sudden peaks in user demand.
However, building and deploying an AI-enabled service at scale for global use is a huge technical challenge that spans model building, training, inference serving and resource scaling. It demands an infrastructure that’s easy to manage and monitor, can deliver real-time performance to end customers wherever they are and can scale as user-demand peaks, all while optimizing costs.
To support their growing user-base, Let’s Enhance chose to deploy their AI-powered platform to production on Google Cloud and NVIDIA. But before diving into their solution, let’s take a close look at the technical challenges they faced in meeting their business goals.
Architecting a solution to fuel the next wave of growth and innovation
To deliver the high-quality enhanced images that end-customers see, Let’s Enhance products are powered by cutting-edge deep neural networks (DNNs) that are both compute- and memory-intensive. While building and training these DNN models in itself is a complex, iterative process, application performance — when processing new user-requests, for instance — is crucial to delivering a quality end user experience and reducing total deployment costs.
A single inference or processing request i.e., from user-generated image, which can vary widely in size, at the input to the AI-enhanced output image, requires combining multiple DNN models within an end-to-end pipeline. The key inference performance metrics to optimize for included latency (the time it takes from providing an input image to the enhanced image being available) and throughput ( the number of images that can be processed per second)..
Together, Google Cloud and NVIDIA technologies provided all the elements that the Let’s Enhance team needed to set themselves up for growth and scale. There were three main components in the solution stack:
- Compute resources: A2 VMs, powered by NVIDIA A100 Tensor Core GPUs
- Infrastructure management: Google Kubernetes Engine (GKE)
- Inference serving: NVIDIA Triton Inference Server
Improved throughput and lower costs with NVIDIA A100 Multi-Instance GPUs (MIG) on Google Cloud
To meet the computational requirements of the DNN models and deliver real-time inference performance to their end-users, Let’s Enhance chose Google Cloud A2 VMs powered by NVIDIA A100 Tensor Core GPUs as their compute infrastructure. A100 GPU’s Multi-Instance GPU (MIG) capability offered them the unique ability to partition a single GPU into two independent instances and simultaneously process two user-requests at a time, making it possible to service a higher volume of user requests while reducing the total costs of deployment.
The A100 MIG instances delivered a 40% average throughput improvement compared to the NVIDIA V100 GPUs, with an increase of up to 80% for certain image enhancement pipelines using the same number of nodes for deployment. With improved performance from the same sized node pools, Let’s Enhance observed a 34% cost savings using MIG-enabled A100 GPUs.
Simplified infrastructure management with GKE
To offer a guaranteed quality-of-service (QoS) to their customers and manage user demand, Let’s Enhance needed to provision, manage and scale underlying compute resources while keeping utilization high and costs low.
GKE offers industry-leading capabilities for training and inference such as support for 15,000 nodes per cluster, auto-provisioning, auto-scaling and various machine types (e.g. CPU, GPU and on-demand, spot), making it the perfect choice for Let’s Enhance.
With support for NVIDIA GPUs and NVIDIA GPU sharing capabilities, GKE can provision multiple A100 MIG instances to process user requests in parallel and maximize utilization. As the compute required for the deployed ML pipelines increases (e.g., a sudden surge in inference requests to service), GKE can automatically scale to additional node-pools with MIG partitions — offering finer granularity to provision the right-sized GPU acceleration for workloads of all sizes.
“Our total averaged throughput varied between 10 and 80 images / sec. Thanks to support for NVIDIA A100 MIG and auto scaling mechanisms in GKE we can now scale that up to 150 images/sec and more, based on user-demand and GPU availability,” said Vlad Pranskevičius, Co-founder and CTO, Let’s Enhance.
In addition, GKE’s support for dynamic scheduling, automated maintenance and upgrades, high availability, job API, customizability and fault tolerance simplified managing a production deployment environment, allowing the Let’s Enhance team to focus on building advanced ML pipelines.
High-performance inference serving with NVIDIA Triton Inference Server
To optimize performance and simplify the deployment of their DNN models onto the GKE-managed node pools of NVIDIA A100 MIG instances, Let’s Enhance chose the open-source NVIDIA Triton Inference Server, which deploys, runs and scales AI models from any framework onto any GPU- or CPU-based infrastructure.
NVIDIA Triton’s support for multiple frameworks enabled the Let’s Enhance team to serve models trained in both TensorFlow and PyTorch, eliminating the need to set up and maintain multiple serving solutions for different framework backends. In addition, Triton’s ensemble model and shared memory features helped maximize performance and minimize data transfer overhead for their end-to-end image processing pipeline, which consists of multiple models and large amounts of raw image data transferring between them.
“We saw over 20% performance improvement with Triton vs. custom inference serving code, and were also able to fix several errors during deployment due to Triton’s self-healing features”, said Vlad. “Triton is packed with cool features, and as I was watching its progress from the very beginning, the pace of product development is just amazing.”
To further enhance end-to-end inference performance, the team is adopting NVIDIA TensorRT, an SDK to optimize trained models for deployment with the highest throughput and lowest latency while preserving the accuracy of predictions.
“With the subset of our models which we converted from Tensorflow to NVIDIA TensorRT we observed a speedup of anywhere between 10% and 42%, depending on the model, which is impressive,” said Vlad. “For memory-intensive models like ours, we were also able to achieve predictable memory consumption, which is another great benefit of using TensorRT, helping prevent any unexpected memory-related issues during production deployment.”
Teamwork makes the dream work
By using Google Cloud and NVIDIA, Let’s Enhance addressed the real-time inference serving and infrastructure management challenges of taking their AI-enabled service to production at scale in a secure Google Cloud infrastructure.
GKE, NVIDIA AI software and A2 VMs powered by NVIDIA A100 GPUs brought together all the elements they needed to deliver the desired user experience and build a solution that can dynamically scale their end-to-end pipelines based on user demand.
The close collaboration with the Google Cloud and NVIDIA team, every step of the way, also helped Let’s Enhance get their solution to market fast. “The Google Cloud and NVIDIA teams are incredible to work with. They are highly professional, always responsive and genuinely care about our success,” said Vlad. “We were able to get technical advice and recommendations directly from Google Cloud product and engineering teams, and also have a channel to share our feedback about Google Cloud products, which was really important to us.”
With a solid foundation and solution architecture that can meet their growing business needs, Let’s Enhance is marching towards their goal of bringing AI-enhanced digital images to everyone. “Our vision is to help businesses effectively manage user-generated content and increase conversion rates with next-gen AI tools,” said Sofi. ”Our next step towards that is to expand our offerings to completely replace manual work for photo preparation, including as well as cover pre-stages as image quality assessment and moderation.”
To learn more about the Let’s Enhance products and services, check out their blog here.
Unlocking Data Value: Latest Data Platforms and Announcements at the Next 21
6418
Of your peers have already read this article.
3:30 Minutes
The most insightful time you'll spend today!
Today at Google Cloud Next we are announcing innovations that will enable data teams to simplify how they work with data and derive value from it faster. These new solutions will help organizations build modern data architectures with real-time analytics to power innovative, mission-critical, data-driven applications.
Too often, even the best minds in data are constrained by ineffective systems and technologies. A recent study showed that only 32% of companies surveyed gained value from their data investments. Previous approaches have resulted in difficult to access, slow, unreliable, complex, and fragmented systems.
At Google Cloud, we are committed to changing this reality by helping customers simplify their approach to data to build their data clouds. Google Cloud’s data platform is simply unmatched for speed, scale, security, and reliability for any size organization with built-in, industry-leading machine learning (ML) and artificial intelligence (AI), and an open standards-based approach.
Vertex AI and data platform services unlock rapid ML modeling
With the launch of Vertex AI in May 2021, we empowered data scientists and engineers to build reliable, standardized AI pipelines that take advantage of the power of Google Cloud’s data pipelines. Today, we are taking this a step further with the launch of Vertex AI Workbench, a unified user experience to build and deploy ML models faster, accelerating time-to-value for data scientists and their organizations. We’ve integrated data engineering capabilities directly into the data science environment, which lets you ingest and analyze data, and deploy and manage ML models, all from a single interface.
Data scientists can now build and train models 5X faster on Vertex AI than on traditional notebooks. This is primarily enabled by integrations across data services (like Dataproc, BigQuery, Dataplex, and Looker), which significantly reduce context switching. The unified experience of Vertex AI let’s data scientists coordinate, transform, secure and monitor Machine Learning Operations (MLOps) from within a single interface, for their long-running, self-improving, and safely-managed AI services.
“As per IDC’s AI StrategiesView 2021, model development duration, scalable deployment, and model management are three of the top five challenges in scaling AI initiatives,” said Ritu Jyoti, Group Vice President, AI and Automation Research Practice at IDC. “Vertex AI Workbench provides a collaborative development environment for the entire ML workflow – connecting data services such as BigQuery and Spark on Google Cloud, to Vertex AI and MLOps services. As such, data scientists and engineers will be able to deploy and manage more models, more easily and quickly, from within one interface.”
Ecommerce company, Wayfair, has transformed its merchandising capabilities with data and AI services. “At Wayfair, data is at the center of our business. With more than 22 million products from more than 16,000 suppliers, the process of helping customers find the exact right item for their needs across our vast ecosystem presents exciting challenges,” said Matt Ferrari, Head of Ad Tech, Customer Intelligence, and Machine Learning; Engineering and Product at Wayfair. “From managing our online catalog and inventory, to building a strong logistics network, to making it easier to share product data with suppliers, we rely on services including BigQuery to ensure that we are able to access high-performance, low-maintenance data at scale. Vertex AI Workbench and Vertex AI Training accelerate our adoption of highly scalable model development and training capabilities.”
BigQuery Omni: Breaking data silos with cross-cloud analytics and governance
Businesses across a variety of industries are choosing Google Cloud to develop their data cloud strategies and better predict business outcomes — BigQuery is a key part of that solution portfolio. To address complex data management across hybrid and multicloud environments, this month we are announcing the general availability of BigQuery Omni, which allows customers to analyze data across Google Cloud, AWS, and Azure. Healthcare provider, Johnson and Johnson was able to combine data in Google Cloud and AWS S3 with BigQuery Omni without needing data to migrate.
This flexible, fully-managed, cross-cloud analytics solution allows you to cost-effectively and securely answer questions and share results from a single pane of glass across your datasets, wherever you are. In addition to these multicloud capabilities, Dataplex will be generally available this quarter to provide an intelligent data fabric that enables you to keep your data distributed while making it securely accessible to all your analytics tools.
Spark on Google Cloud simplifies data engineering
To help make data engineering even easier, we are announcing the general availability of Spark on Google Cloud, the world’s first autoscaling and serverless Spark service for the Google Cloud data platform. This allows data engineers, data scientists, and data analysts to use Spark from their preferred interfaces without data replication or custom integrations. Using this capability, developers can write applications and pipelines that autoscale without any manual infrastructure provisioning or tuning. This new service makes Spark a first class citizen on Google Cloud, and enables customers to get started in seconds and scale infinitely, regardless if you start in BigQuery, Dataproc, Dataplex, or Vertex AI.
Spanner meets PostgreSQL: global, relational scale with a popular interface
We’re continuing to make Cloud Spanner, our fully managed, globally scalable, relational database, available to more customers now with a PostgreSQL interface, now in preview. With this new PostgreSQL interface, enterprises can take advantage of Spanner’s unmatched global scale, 99.999% availability, and strong consistency using skills and tools from the popular PostgreSQL ecosystem.
This interface supports Spanner’s rich feature set that uses the most popular PostgreSQL data types and SQL features to reduce the barrier to entry for building transformational applications. Using the tools and skills they already have, developer teams gain flexibility and peace of mind because the schemas and queries they build against the PostgreSQL interface can be easily ported to another Postgres environment. Complete this form to request access to the preview.
Our commitment to the PostgreSQL ecosystem has been long standing. Customers choose Cloud SQL for the flexibility to run PostgreSQL, MySQL and SQL Server workloads. Cloud SQL provides a rich extension collection, configuration flags, and open ecosystem, without the hassle of database provisioning, storage capacity management, or other time-consuming tasks.
Auto Trader has migrated approximately 65% of their Oracle footprint to Cloud SQL, which remains a strategic priority for the company. Using Cloud SQL, BigQuery, and Looker to facilitate access to data for their users, and with Cloud SQL’s fully managed services, Auto Trader’s release cadence has improved by over 140% (year-over-year), enabling an impressive peak of 458 releases to production in a single day.
Looker integrations make augmented analytics a reality
We are announcing a new integration between Tableau and Looker that will allow customers to operationalize analytics and more effectively scale their deployments with trusted, real-time data, and less maintenance for developers and administrators. Tableau customers will soon be able to leverage Looker’s semantic model, enabling new levels of data governance while democratizing access to data. They will also be able to pair their enterprise semantic layer with Tableau’s leading analytics platform. The future might be uncertain, but together with our partners we can help you plan for it.
We remain committed to developing new ways to help organizations go beyond traditional business intelligence with Looker. In addition to innovating within Looker, we’re continuing to integrate within other parts of Google Cloud. Today, we are sharing new ways to help customers deliver trusted data experiences and leverage augmented analytics to take intelligent action.
First, we’re enabling you to democratize access to trusted data in tools where you are already familiar. Connected Sheets already allows you to interactively explore BigQuery data in a familiar spreadsheet interface and will soon be able to leverage the governed data and business metrics in Looker’s semantic model. It will be available in preview by the end of this year.
Another integration we’re announcing is Looker’s Solution for Contact Center AI, which helps you gain a deeper understanding and appreciation of your customers’ full journey by unlocking insights from all of your company’s first-party data, such as contextualizing support calls to make sure your most valuable customers receive the best service.
We’re also sharing the new Looker Block for Healthcare NLP API, which provides simplified access to intelligent insights from unstructured medical text. Compatible with Fast Healthcare Interoperability Resources (FHIR), healthcare providers, payers, and pharma companies can quickly understand the context and relationships of medical concepts within the text, and in turn, can begin to link this to other clinical data sources for additional AI and ML actions.
Bringing the best of Google together with Google Earth Engine and Google Cloud
We are thrilled to announce the preview of Google Earth Engine on Google Cloud. This launch makes Google Earth Engine’s 50+ petabyte catalog of satellite imagery and geospatial data sets available for planetary-scale analysis. Google Cloud customers will be able to integrate Earth Engine with BigQuery, Google Cloud’s ML technologies, and Google Maps Platform. This gives data teams a way to better understand how the world is changing and what actions they can take — from sustainable sourcing, to saving energy and materials costs, to understanding business risks, to serving new customer needs.
For over a decade, Earth Engine has supported the work of researchers and NGOs from around the world, and this new integration brings the best of Google and Google Cloud together to empower enterprises to create a sustainable future for our planet and for your business.
At Google Cloud, we are deeply grateful to work with companies of all sizes, and across industries, to build their data clouds. Join my keynote session to hear how organizations are leveraging the full power of data, from databases to analytics that support decision making to AI and ML that predict and automate the future. We’ll also highlight our latest product innovations for BigQuery, Spanner, Looker, and Vertex AI.
I can’t wait to hear how you will turn data into intelligence and look forward to connecting with you.
What Swiggy and You Can Learn From This Company’s Use of ML to Engage Customers
9639
Of your peers have already read this article.
1:45 Minutes
The most insightful time you'll spend today!
The app economy has enabled a huge range of unique business models to flourish. One such model is online food ordering and delivery services, in which apps leverage geo-location data to aggregate local food choices and offer personalized options to consumers.
A leading company in this space is Just Eat. Launched in the UK in 2001 with a vision of ‘serving the world’s greatest menu. Brilliantly.’ The company has capitalized on the popularity of online food delivery and grown its presence across 12 markets.
Just Eat acts as an intermediary between take-out food outlets and hungry customers, giving local restaurants access to a broader base of potential diners, while providing consumers with an easy and secure way to order and pay for food from their favourite restaurants.
Today the company helps 27 million customers find food from more than 112,000 restaurants—everything from homemade Italian pasta, to Chinese noodle bowls, to fish-and-chips.
Data is the fuel of Just Eat’s rapid growth, but it wasn’t always looked at that way. In its early days, Just Eat struggled with the deluge of information and faced fragmentation across its systems. In fact, the company realized its legacy data vendor wasn’t capable of ingesting 90 percent of the data produced by its food platform. This was incredibly frustrating for Just Eat’s analysts and data scientists, who had to waste time cleaning up sources instead of leveraging the data to create a better user experience.
Just Eat turned to Google Cloud, and now uses machine learning (ML) to power sophisticated consumer recommendations on both its app and website. It also makes heavy use of features offered by Google Cloud Platform, including BigQuery for running analytics on its customer data set and Cloud Pub/Sub for messaging app users with relevant offers in real-time.
Having all of Just Eat’s data in one platform has translated into real value for its customers. With Google Cloud tools, Just Eat has created its own proprietary Customer Ontology framework, which today contains 5.5 billion features that better understand consumers’ behavior and food habits, and provides insights into previous visits.
Just Eat recently created an “Adventurous Index” to map its customers according to their ordering habits, enabling them to tailor their marketing and user experiences. For example, mid-adventurous customers are shown a choice of restaurants that serve their most ordered cuisine, while adventurous customers can choose from restaurants that serve a wider variety. This not only has prompted consumers to be more adventurous with their choices, but also has led to more business at a more diverse set of restaurants.
Matt Cresswell, Director of Customer Platforms at Just Eat said that Google Cloud has become integral to its product delivery: “Consumer food choice is a hugely nuanced topic. We know that individuals have their own unique journeys when they use Just Eat. We’ve sought to create a truly one-to-one relationship with every customer. The changes we’ve made to the platform mean they can access the dishes they enjoy at the touch of a fingertip, and find inspiration to discover new dishes they’ll love. We’re grateful to Google Cloud for helping us support our customers on their culinary explorations.”
More Relevant Stories for Your Company
Master AI Prompt Engineering with 6 Proven Tips
As AI-powered tools become increasingly prevalent, prompt engineering is becoming a skill that developers need to master. Large language models (LLMs) and other generative foundation models require contextual, specific, and tailored natural language instructions to generate the desired output. This means that developers need to write prompts that are clear,
Engage and Translate: Text and Audio Chat in 100+ Languages
As users worldwide connect to the internet and work remotely, it’s more important than ever to make interactive applications chat in many languages. But when users speak 100s of languages, this quickly can get challenging. Where to start? If you are new to translation, Sarah Weldon, the Product Manager for
The 5-Min Demo: How to Develop, Train and Deploy Training Models on Kubernetes
Machine learning has taken businesses by storm. A growing number of organizations, both large and small, and spread across a swathe of industries, are figuring out how to quickly adopt ML. But in the midst of that accelerated push, infrastructure and data science teams have to come to terms with
An Expert’s Opinion on What Early-stage Startups Must Know
As lead for analytics and AI solutions at Google Cloud, my team works with startups building on Google Cloud. This puts us in the fortunate position to learn from founders and engineers about how early-stage startups’ investments can either constrain them or position them for success, even at the seed
