Architecting AI Solutions in Azure for Business
0 likes75 views
The document outlines the architecture and implementation of a hybrid AI-based recommender system using Azure, detailing methods for user profiling, behavioral analysis, and collaborative filtering. It emphasizes the use of unique user identifiers for tracking behaviors, data collection for model training, and continuous improvement of recommendation algorithms. Additionally, it discusses integration with various Azure services for AI applications, such as image recognition in an AI-enabled cash register system.
Architecting AI Solutions in Azure for Business
- 1. Architecting AI Solutions in Azure for Business From Concept to Reality
- 4. Uniquely Identify Users • User has UniqueID regardless of the fact whether they are registered or not. The UniqueID for unregistered user could be assigned based. Below are a couple of ways of generating such unique ID BrowserID • Unique id generated from the browsers user agent string. Browser|BrowserVersion|OS|OSVersion|Processor|MozzilaMajorVersion|GeckoMajorVersion ComputerID • Generated from users IP Address and HTTPS session key. getISP(requestIP)|getHTTPSClientKey() FingerPrintID • JavaScript based fingerprinting based on a modified fingerprint.js.FingerPrint.get() SessionID • Random key generated when user 1st visits site. BrowserID|ComputerID|randombytes(256) GoogleID • Generated from __utma cookie. getCookie(__utma).uniqueid Unregistered users • generate uniqueID as above and store it in a cookie. • User actions in the website will be traced: Searches; Views; Likes; View Contact Information
- 5. Functional Features User Profiling • Establish detailed user profiles by collecting information such as industry, preferences, recent searches, and location. • Encourage users to input additional preferences and requirements to fine-tune recommendations. Contextual Awareness • Utilize geolocation data to offer services relevant to the user's location. • Question: Do we have time-sensitive offers, expiring during day or limited time? Consider temporal factors, such as time of day and day of the week. Behavioral Analysis • Analyze user behaviour, including clicks, time spent on offer pages and even offer previews • Analyze behaviour of interactions (likes, comments).
- 6. Functional Features (Continued…) Collaborative Filtering • Implement collaborative filtering to reflect the preferences and feedback of similar users. Supplier Score • Integrate a supplier scoring system based on user reviews, ratings and purchase history. Suppliers with high scores should be considered reputable and reliable service providers. Real-time Feedback • Allow users to provide feedback on recommended services directly through the banner. I.e. “hide” and “add to favorites” with sub options to hide this offer or offers from that supplier • The feedback shall be used to continuously improve the recommendation algorithm output for the user, but also to affect the collaborative score. Transparency • Provide user with transparency on why certain services are recommended by highlighting key factors such as user preferences, location relevance, and supplier scores. (i.e. decomposing score) • The point will also be useful for assessment during development time
- 7. Types of Feature Filters Content-based filtering (similarity between item features) • The underlying assumption of the collaborative filtering approach is that if a person A has the same opinion as a person B on a set of items, A is more likely to have B's opinion for a given item than that of a randomly chosen person. • Method makes automatic predictions (filtering) about the interests of a user by collecting preferences or taste information from many users (collaborating). Collaborative item filtering (similarity between items based on interactions) • Uses only information about the description and attributes of the items users has previously consumed to model user's preferences. • Algorithms tries to recommend items that are similar to those that a user liked in the past (or is examining in the present). • Various candidate items are compared with items previously rated by the user and the best-matching items are recommended.
- 8. Hybrid Feature Filters Hybrid Filter Recommender • Combining collaborative filtering and content-based filtering could be more effective than pure approaches in some cases. • These methods can also be used to overcome some of the common problems in recommender systems such as cold start (no previous history). RecommenderHybridScore = RecommenderContentScore*ContentWeight + RecommenderCollaborativeScore*ContentWeight
- 9. Feature Types # Feature Guest User Weight Auth User Weight Type F01 Location 5 5 Content F02 Trending offers (offer views and interactions score) 5 2 Collaborative F03 Person/Organization keywords, profile N/A 10 Collaborative F04 View time (time spent on offer during scrolling) (most recent = most important) 3 Collaborative F05 Search history, Type of service 7 7 Collaborative F06 Promoted offers (a paid service) 10 5 Content F07 Content (keywords, service type) 10 10 Content F08 Clicks (most recent = most important) 15 10 Collaborative F09 Interactions (likes, comments, views of contact details) (most recent = most important) N/A 20 Collaborative F10 Realtime feedback – hide offer, hide supplier 20 30 Collaborative
- 10. Recommender Flow Identification • Unique User ID - Used for tracking user behaviour on the web site • Anonymous users - Automatically generated from the browser agent and IP location • Authorized users Incentivisation • The web portal shall support SSO using both Facebook and Google to increase the likelihood to identify users • There shall be certain hidden information when accessing an offer to urge the end user to log in. (i.e. contact info visible only for registered users, although for free). Data Collection • User-Based Collaborative Data • Anonymous user, Authenticated user • Authenticated user only • History of purchases • History of visited offers (with more weight on the most recent offers accessed) • History of interactions
- 11. Recommender Flow (Continued…) • Item-Based Collaborative Data • User preferences, behavior, and interactions with the platform - historical ratings, reviews, and any explicit feedback. • Time spent reading an offer. Look at Facebook and Instagram reels, as well as TikTok. • The offer feed may be different for each user and could be implemented with infinite scroll where the user navigates through a summary view (images changing in a slideshow, title, short description) and the time spent to read the summary is used as a feature indicating the level of interest of end user • Content-based Data • Information about the items or services being recommended - attributes, features, or content-related information about each item. Data Preprocessing • Clean and preprocess the data. Handle missing values, remove duplicates, and format the data for easy processing User-Based Collaborative Filtering o Calculate user-based collaborative filtering score o Build a user-item matrix: Create a matrix where rows represent users, columns represent items, and the cells contain ratings or interactions. o Calculate user similarities: Use metrics like cosine similarity or Pearson correlation to measure the similarity between users based on their preferences. o Predict ratings for items: Predict the ratings for items that a user has not interacted with based on the ratings of similar users.
- 12. Recommender Flow (Continued…) Item-Based Collaborative Filtering • Transpose the user-item matrix: Create an item-user matrix by transposing the user-item matrix. • Calculate item similarities: Measure the similarity between items based on user interactions using techniques like cosine similarity. • Predict ratings for users: Predict the ratings a user might give to items they haven't interacted with based on the ratings of similar items. Content Filtering • Extract features for items: For each item, extract relevant features or attributes. This could include textual data, categorical data, or any other relevant information. • Build an item-feature matrix where rows represent items, columns represent features, and the cells contain the values of the features. • Calculate content similarities. Use similarity measures (e.g., cosine similarity) to determine how similar items are based on their features. • Combine collaborative and content similarities. This could involve weighting each similarity measure based on its importance. Hybrid Recommender • Combine the recommendations from collaborative and content filtering, giving appropriate weight to each.
- 13. Evaluation and Improvement Evaluation • Split the data into training and testing sets. Evaluate the performance of your recommender system by comparing predicted ratings or recommendations with actual user interactions in a testing set. • Utilize metrics such as Root Mean Squared Error (RMSE) – i.e. the distance between real and predicted score Continuous Improvement • Implement mechanisms to continuously update and retrain the recommender system as new data becomes available. Takeaways • Content Based Recommendation System o https://www.youtube.com/watch?v=9_YdIjGl2m4 • Kaggle Music Recommender o https://www.kaggle.com/code/vatsalmavani/music-recommendation-system-using-spotify-dataset • Kaggle Movie Recommender o https://www.kaggle.com/code/rounakbanik/movie-recommender-systems • Kaggle E-commerce Recommender o https://www.kaggle.com/code/shawamar/product-recommendation-system-for-e-commerce • Kaggle Content based Recommender o https://www.kaggle.com/code/omeroruccelik/content-based-recommendation-systems • Recommender System in Python o https://www.kaggle.com/code/gspmoreira/recommender-systems-in-python-101 • Azure ML Recommender System o https://azure.microsoft.com/en-us/blog/building-recommender-systems-with-azure-machine-learning-service/
- 14. App Service - Hybrid Connection • Access on-premises AZ SQL DB from the cloud • Install Hybrid Connection Manager on-premises
- 16. AI-Enabled Cash Register • Snap a photo with food item • Device detects items from photo • App determines item attributes
- 17. Azure Service – to – Function Mapping Function App • Receives images for scoring (Flow 1) • Receives prediction to be saved (Flow 2) • Synchronizes device config. with DB AKS • Hosting AI Models AZ ML Service • Labeling images • Datasets • Models • Training • Deployments Function App (Builder) • Builds datasets • Crates ML job • Deploys models Web App • Show prediction results • Manage known items and daily tasks (menu) Azure SQL DB • Asset Structure • Menus • Predictions IoT Hub • Device Configuration Storage • Store prediction images
- 18. Gateway Module Architecture Flow Camera captures image Deduplication (Scene has Changes) Check Target - Crop image to tray area Deduplication - Check image focus Check for Known Items (Flow 1) Deduplication - Check items on tray GDPR Check – Personal items Upload image to storage Persist Image (Flow 2) Tray Detection AI Model GDPR Detection AI Model
- 19. Auxiliary Flows RPI Device API Management Dynamic Endpoint AZ ML Deployment Endpoint RPI Device Azure Function Dynamic Endpoint Flow 2: Saving Prediction Results Flow 1: Send Items for Object Detection
- 20. Object Scene Deduplication • Deduplication 1: Image must be at least 20% different that the previous one. Comparing image hash values. • Deduplication 2: Image must pass blurriness threshold. Checking pixel variance. • Deduplication 3: Comparing polygons (fingerprint) made of the centroids of the detected foods between two images.
- 21. GDPR Model
- 22. Model Application (Good Results)
- 23. Detection Issues from Labeling • Too small polygon shape • Too simple polygon shape • Too complex polygon shape • Wrong polygon shape • ML assisted labeling glitch
- 24. Detection Issues from Labeling • Too small polygon shape • Too simple polygon shape • Too complex polygon shape • Wrong polygon shape • ML assisted labeling glitch
- 25. Detection Issues from Labeling • Too small polygon shape • Too simple polygon shape • Too complex polygon shape • Wrong polygon shape • ML assisted labeling glitch
- 26. Detection Issues from Labeling • Too small polygon shape • Too simple polygon shape • Too complex polygon shape • Wrong polygon shape • ML assisted labeling glitch
- 27. Detection Issues from Labeling • Too small polygon shape • Too simple polygon shape • Too complex polygon shape • Wrong polygon shape • ML assisted labeling glitch