MIMI AI ENGINE
Foundational multisensory AI technology serving as the building block of AI transformation
Entefy Mimi AI represents the next-generation context-aware machine intelligence technology. It’s comprised of a set of modular AI models and intelligent orchestration frameworks that work together to extract meaningful insights from diverse and massively multivariate datasets. Mimi AI models can be deployed in virtually endless combinations to provide critical AI functionality to practically any system or operation an organization envisions.
Mimi AI Engine operates at the heart of the Entefy 5-Layer Platform and incorporates an integrated set of advanced AI technologies including MimiVision, MimiLanguage, MimiCognition, MimiAudio, and MimiData.
MimiVision
Machine vision measured in billions of parameters
A picture is worth far more than 1,000 words. MimiVision’s computer vision technology involves the training of multiple deep networks composed of many narrow and wide layers. Each training requires the estimation of approximately 1 billion parameters/coefficients with each coefficient roughly corresponding to a neuron. The final ensemble of models is optimized and fine-tuned for detecting objects, attributes, and relationships from images and videos.
Key Features
- Object classification
- Object detection
- Scene detection and analysis
- Facial recognition
- Facial analysis (including emotion, age, gender)
- Contextualization
- Similarity-based retrieval
- Object-to-object mapping and relationships
- Object tracking in video
- Color identification
- Depth, focus, and saliency
- Logo detection
- Celebrity recognition
- Places of interest
- Optical character recognition (OCR)
MimiLanguage
Context-aware natural language intelligence
Entefy’s natural language processing (NLP) and natural language understanding (NLU) capabilities leverage deep learning and other machine learning methods to enable advanced AI that detects meaning and intent in unstructured data. This enables a multitude of valuable capabilities including entity recognition, information extraction, intent analysis, and summarization, all of which increase the value and discoverability of text-based data.
Key Features
- Part-of-speech tagging
- Semantic parsing
- Spelling and grammar correction
- Q&A engine
- Topic modeling
- Entity extraction
- Relationship extraction
- Sentiment analysis
- Topic segmentation
- Word sense disambiguation
- Intent analysis
- Concept mapping
- Context mapping
- Machine summarization
- High-speed concept mapping
MimiCognition
Cognitive machine intelligence for understanding, inference, learning, and engagement
Inspired by human reasoning and understanding, smart machines are becoming smarter. Much smarter. Entefy’s cognitive system incorporates multiple machine learning models in computer vision, natural language, audio, time series, and other data intelligence to better understand concepts and contexts. MimiCognition is designed for higher-level cognitive tasks for machines.
Key Features
- Intelligence, insights, and actions
- Multimodal knowledge extraction
- High-frequency event correlation analysis
- Cross-modal entity recognition and understanding
- Dynamic intent analysis for workflow integration
- Real-time intent-context mapping and execution
- Support for user-specific context learning and intent management
- Batch and online learning capable
- Broad support for structured and unstructured data
- Seamless integration with MimiVision, MimiLanguage, MimiAudio, and MimiData
- Granular data access controls and policy management capabilities
- Support for secure enclave and compute sandbox implementations
- Full data storage integration with internal and external repositories
- Support for Mimi SmartAgent
MimiAudio
Audio intelligence to extract meaning from audio signals
Enable valuable insights from audio. With Entefy’s machine learning models for audio stream processing, users can make use of audio information embedded in recordings, presentations, videos, user-generated content, and more. This translates to new capabilities for your team and better intelligence for you and your organization.
Key Features
- Audio information retrieval
- Audio source separation
- Classification of audio streams for speech, music, and other categories
- Speech synthesis
- Speech transcription
- Speaker detection
- Music transcription
- Sentiment analysis
- Audio disambiguation
- Speech generation
- Music genre classification and tagging
- Audio file matching and similarity search
MimiData
Descriptive and prescriptive data intelligence for time series and other tabular data
Understand the past, the present, and know what to expect in the future. Entefy’s data intelligence models transform structured and unstructured data into meaning. Important correlations, anomalies, and predictions are just a few of the many advanced insights that can be derived from smart sensors, network activity, global news, health systems, financial transactions, and other sources of data.
Key Features
- Descriptive, predictive, and prescriptive analytics
- Univariate and multivariate time series forecasting
- Simultaneous analysis of disparate data sources
- Automated management of data duplication and stale data
- Health and failure analysis and prediction
- Batch and real-time analysis of patterns, anomalies, and concepts
- Data intelligence support for Mimi SmartAgent
- Automatic risk and KPI analysis
- Support for batch and online learning
- Optimization modules for performance and processes
- Seamless integration available for MimiVision, MimiLanguage, and MimiAudio
- Support for diverse data streams from third-party services, apps, and IoT smart devices
- Insights to trigger notifications and other actions
