If any solution requires intervention of AI and ML then there are two ways to go about it:

1. Creating Core Competency: This methods needs creating AI and ML product by scratch.

   PRO(s)	CON(s)
   You have total control on your dataset and privacy	You need too much data to have good training.
   You can experiment with new features.	Labelling existing data to train model is challenge.
   You can improve existing feature via alternate Models	Start time of project is high
   Best suited for companies which can do large investments.	Initial cost is high in terms of Man power and Server Infrastructure

2. Using AI and ML as Service (MLaaS): There are multiple providers in market who have already trained their models.

   PRO(s)	CON(s)
   Ready made and Generic ready to use array of APIs. Pay as you use.	You have partial control on your dataset and privacy.
   Already Trained and Defined model.(By Google/ AWS/ MS/ Others)	User is limited by functionality of service provider.
   Projects can start quickly over requirement and use cases.	User is limited by correctness of service provider.
   Resource cost is low. No infrastructure/ computation to maintain.

MLaaS – How It Works

• Arriving to Use Cases, Solution Provider and Required Output.
  Organization defines the Problem Statement

  Identify multiple MLaaS Use Case as per problem statement

  Lookout for Use Case Solutions across multiple MLaaS providers

  Combine multiple solutions to provide Unified Output
• High Level Structure Diagram and Interactions.

MLaaS – Service Providers

There are multiple service providers and they provide very unique value proposition in following terms:

1. Services and their Support.
2. Features (Quantitative) and Depth in features (Qualitative).
3. Pricing
4. Pre solved Use cases
5. Additional Infrastructure support.

Few of the very prominent service providers are:

1. Google - https://cloud.google.com/ml-engine/
2. Amazon AWS - https://aws.amazon.com/machine-learning/
3. Microsoft - https://azure.microsoft.com/en-us/services/machine-learning-studio/
4. IBM - https://www.ibm.com/cloud/machine-learning
5. BigML - https://bigml.com
6. MachineBox - https://machinebox.io/

MLaaS – Use Cases

MLaaS has already seen uses across various industries. It is being used in processes such as risk analytics, fraud detection, manufacturing, supply chain optimization, network analytics, marketing, advertising, predictive maintenance, and inventory management optimization, among others.

The application spans across various industries as well, such as healthcare, banking, financial services and insurance (BFSI), transportation, retail, manufacturing, and telecom, among others.

Most prominent use cases are listed as:

• Video Intelligence:
  • Automatically recognize a vast number of objects, places, and actions in stored and streaming video.
  • Also identifies text spoken in video. Easily search your video catalog the same way you search text documents.
• Vision Intelligence:
  • Face Detection and Face Attribute Detection
  • Emotion Detection
  • Facial Recognition
  • Person detection
  • Liveness
  • Heat- mapping
  • Tracking and merging offline/online experiences
  • Employee management
  • Biometric surveillance
  • Fraud Prevention and I.D. Unauthorized Visitors
  • Customer Service and Improving Scoring
  • Using Safe search, and estimate the likelihood that any given image includes adult content, violence, and more.
• Natural Language Processing:
  • Sentiment analysis
  • Entity Recognition
  • Document and content classification
  • Document (receipt, invoice and others) understanding
  • Automated document processing workflow
  • Make physical document quickly searchable
• Translation:
  • Translate between languages
  • Language detection
  • Cover >50 languages
• Cloud Speech-to-Text API
• Cloud Text-to-Speech API
• Chatbots:
  • Conduct rich, natural interactions with automatic spelling correction
  • Sentiment Analysis
  • Recognize Multiple Speakers
  • Channel Identification
  • Deliver automated phone service
  • Recognize voice interactions and generate a voice response
  • Bulk add data from your enterprise to your agent
• Analytics
  • Analyzing foot traffic and conversion for retailers
  • Detecting data anomalies.
  • Identifying correlations in real time over sensor data, or generating high-quality recommendations.
  • Analytics dashboard.
• Forecast and Recommendations
  • Forecast can use virtually any historical time series data (e.g., price, promotions, economic performance metrics) to create accurate forecasts for your business.
  • For example, in a retail scenario, Forecast uses machine learning to process your time series data (such as price, promotions, and store traffic) and combines that with associated data (such as product features, floor placement, and store locations) to determine the complex relationships between them.

MLaaS – What we have done

Team at Fastcurve Services has used Google cloud and AWS MLaaS APIs very extensively. Team has prepared pre-release version/ prototype/ proposal of following use cases for clients (Yet to be phased out in production) :

• Chatbots (for Automated support)
• Online and Discoverable Documents with attributes and classification (converted from Offline and Physical Documents)
• Face Recognition for Authentication, Automated identification and Reporting.
• Sensor Malfunctioning Alarm integration, reporting and auto assignment of task for fix.