AI projects

Proof of concept of the state of the art AI with practical & research examples (with code demos)

• narrow AI
• general AI
• super AI

Collections of algorithms optimization

• Optimal path using BFS, DFS
• AI search algo

AI search algo

• dijkstra search
• heuristics
• A* algo

AI algorithm

• Determined the optimal next move of a chessboard game using Minimax algorithm with Alpha-beta pruning

• The minimum cost transaction for a goal state
• A sequence of transitions to a minimum cost goal
• A minimum cost transaction for a minimum cost goal

AI in Finance

AI in Bioinformatics

AI in game

• rule based system

• Prolog

  • swipl
  • brew install swi-prolog
  • prolog query

• The min-max algorithm

Edge AI

• edge service
  • smartphone
  • devices
  • microcontroller
• openvm
• jevois
• google edge TPU
• movidius
• nvidia jetson
• UP AI Edge
• Ultra96
• TF Lite
• utensor
• qualcomm neural processing SDK for AI
• huawei NPU

AI use case

• fraud detection

• https://www.gurobi.com/

• integer linearn programming

  • https://www.gurobi.com/
  • https://www.ibm.com/analytics/cplex-optimizer

• robotics

  • https://www.ros.org/

AI in IIoT

• optimize logistics

• Electrical load forecasting

• Implementing a code to perform preventive maintenance based on aircraft engine sensors data

• deploy machine-to-machine (M2M) and machine-to-human (M2H) communication, along with AI-powered analytical algorithms, enabling predictive maintenance, that predict the breakdown before it occurs using past data.

• monitoring parameters/sensor

  • Vibration sensors mainly used to detect misalignment, imbalance, mechanical looseness, or wear on pumps and motors
  • Current/voltage sensors to measure the current and voltage supplied to an electric motor
  • Ultrasound analysis to detect leakage in pipe systems or tanks, or mechanical malfunctions of movable parts and faults in electrical equipment
  • Infrared thermography to identify temperature fluctuations
  • Sensors to detect liquid quality (for example in the case of wine sensors to detect the presence of different elements in the wine)

• DL model: RNN, LSTM

• STLF using LSTM

  • dataset : https://archive.ics.uci.edu/ml/datasets/Individual+household+electric+power+consumption#
  • 2 LSTM and 1 connected layer

AI in Cybersecurity

• Predictive model for credit card fraud detection

  • big data analytics to integrate information from different sources
  • ensemble learning
    • Use bagging and boosting algorithms
    • Adaptive Boosting (AdaBoost)
    • gradient boosting algorithm
  • sampling techniques to rebalance datasets, thereby improving prediction accuracy
    • Oversampling with SMOTE
      • Synthetic Minority Over-sampling Technique (SMOTE)

• GANs - Attacks and defense

  • forward propagation
  • backpropagation

• Feedforward neural networks (FFNNs)

• Recurrent neural network (RNNs)

  • network traffic analysis

• Convolutional neural networks (CNNs)

• Spam detection

• Fraud detection algorithms

• Biometric authentication with facial recognition

• Classifying suspicious user activity

• User authentication with keystroke recognition

• Suspect fraud

• Application security :

  • attacks : SSRF, SQL injection, XSS, DDoS

• Endpoint protection

  • ransomware

• Network protection

  • intrusion detection system

• Some tasks

  • Predict : NN, DL
  • Clustering

• Multi Layer Perceptron

• Using :

  • https://imbalanced-learn.readthedocs.io/en/stable/generated/imblearn.under_sampling.RandomUnderSampler.html

AI in IoT

• Self driving solution

• Safe route parameter to trip planners

• Apply CNN to parking lot

• Apply SVM to safety on trip planning

• Teaching MDP to find the safest route

• Perform supervised and unsupervised machine learning for IoT data

• Implement distributed processing of IoT data over Apache Spark using the MLLib and H2O.ai platforms

• Forecast time-series data using deep learning methods

• build smart systems for IoT

• monitor heart disease using ML

• Smart home

• devices used in smart home

• AI in predicting human activity recognition

• set up RL-DL-CRLMM model

  • webcam images in real time
  • CRL- CNN
    • gap in parking lot
    • SVM - optimizer
    • MDP
    • RL - DL - CRLMM find parking lot - available space
    • Circular RL- DL - CRLMM
      • CNN
      • Markov decision process MDP
      • CRLMM - recognize parkigng space in parking lot and send signal to self signal to self driving vehicle
        • gaps, space between 2 objects
        • context to establish whether this space between objects is positive or negative distance

• IP camera : obtain right real time frames from webcam : lighting const, etc

• Dataset :

  • training set, test set

• model trained : CNN Concept Strategy. py

  • Classify parking lot :

• Add SVM function to increase safety level

  • avoid traffic
  • read lat/long of datapoint in another table to convet back to GPS format
  • sklearn
  • make_blobs

• classify

• IP camera

  • Webcam can be tested
  • webcam freeze a frame of a parking lot

• Computer vision

  • simulate frozen frame

• Run CRLMM

  • Find parking space
  • CRL-MM-IoT-SVM.py

• decide how to get to the parking lot

  • crlmm == 1
  • find a safe route to SDC -> activate SVM -> safeSVM() -> traffic graph
  • send info to Google Maps -> script to read dataset that contains GPS coordinate for each datapoint in the SVM

• Itinerary graph

• Weight vector

  • vertex weights (safest route) are updated after MDP

• AI in heath care

  • Heart_Disease_Prediction
    • dataset : https://archive.ics.uci.edu/ml/datasets/heart+Disease
      • 76 attributes
    • SVC classifier & experiment with MLP classifier

AI in Robotics

Data Access and Distributed Processing for IoT

• Hadoop's Distributed File System

• HDF5

  • PyArrow's filesystem interface for HDFS

• SQL, NoSQL

• Dataset

  • 9,568 data points collected from a combined cycle power plant (CCPP)
    • http://archive.ics.uci.edu/ml/datasets/combined+cycle+power+plant
  • Wine quality dataset
    • https://archive.ics.uci.edu/ml/datasets/Wine+Quality
  • Air quality data
    • https://www.kaggle.com/c/predict-impact-of-air-quality-on-death-rates

ML algorithms

Image classification

• Build Nearest neighbour classifier for classifying different categories of images using K Means Clustering for effiency
• Component analysis - histogram
• Classification feature
• Different distance measures for the nearest neighbour classifier was evaluated

Recommendation system

• Cluster algorithm - Reduce search space
• MapReduce to process large dataset
• ML model designed for content-based recommendation
• Cluster algorithm - reduce search space
• Leverage locality sensitive hashing LSH method to find similar users for a large dataset - 1GB

• BM25 weighting
• Efficient nearest neighbor search
• matrix factorization
• https://github.com/benfred/implicit
• efficient nearest neighbor search: https://github.com/facebookresearch/faiss

Image drawer program Mona Lisa

Deep learning

• backpropagation
• gradient descent
• “skip connections”
• batch normalization
• RNN : text, speech , time series data
• XOR
• multi layer, feed forward NN

Reinforcement learning

• Building a learning agent
• RL algorithms
  • Markov process Hidden Markov Models (HMM)
  • Q Learning
  • Temporal difference methods
  • Monte Carlo methods

NLP & sentimental analysis with RNTN

• Background on natural language processing (NLP) and sentiment analysis

• Core NLP: https://stanfordnlp.github.io/CoreNLP/

  • NLP processing such as sentence detection
  • word detection
  • part-of-speech tagging, named-entity recognition (finding names of people, places, dates, and so on), and sentiment analysis.
  • Several NLP features, such as sentiment analysis, depend on prior processing including sentence detection, word detection, and part-of-speech tagging.
  • 85.4% accuracy for detecting positive/negative sentiment of sentences.

• Recursive neural tensor networks (RNTN)

• twitter & reddit api

• Data aggregation

• Sentiment detector

  • libraries, hbc-core, JRAW, and Crux.

• Speech Recognizer

• transform audio signal

• generate audio signal

• synthesizing tones to generate music

• extract speech features

• recognize spoken words with Hidden Markov Model

CoreNLP processing pipeline

• tokenization
• dependency tree
• annotations
• part-of-speech tags

Optimize running time

• Parallel processing and fault tolerance

• Optimize Map Reduce framework

  • Support parallel processing
  • Optimize scripts for map and reduce stage

• Distributed

Google Cloud AI Services

• Cloud based machine learning

• Cloud Vision API

  • detect explicit content
  • landmark detection
  • optical character recognition
  • face detection
  • image attributes

• Cloud Speech API

• Cloud AutoML

• Cloud TPU

• Cloud ML engine

• Cloud natural language

  • syntax analysis
  • entity recognition
  • sentiment analysis
  • multi language
  • integrated REST API

• Cloud Speech API

  • global vocab
  • streaming recognition
  • word hints
  • real time / prerecorded audio support
  • noise robustness
  • inappropriate content filtering

• cloud translation API

• cloud vidio inteligence

  • label detection
  • shot change detection
  • video trans
  • explicit content detection

project detection-gcloudvision

• face detection

• label detection

• safe search detection

• video inteligence api

  • label, search video catalogues, distinguish scenes using shot detection
  • content recommendation, content moderation, contextual ads, search media archives

• cloud speech api

  • streaming speech recognition
  • audio to text with speech recognition

• cloud NLP

  • sentiment analysis
  • entity analysis

Tech stack

• Java

  • NN : (http://neuroph.sourceforge.net/index.html), Deeplearning4j
  • NLP : CoreNLP, OpenNLP
  • ML : JavaML, Weka, SMILE
  • ComputerVision : JavaCV
  • Tensorflow
    • on spark
    • SparkDL
    • PySpark
    • keras

• OpenAI gym

• Open CV

• Python

• Prolog

• https://github.com/spotify/snakebite

• HDFS

Resources / Ref

• https://www.cs.waikato.ac.nz/ml/weka/

• https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/burstable-performance-instances.html
• https://cloud.google.com/products/
• https://blogs.nvidia.com/blog/2016/07/29/whats-difference-artificial-intelligence-machine-learning-deep-learning-ai/
• http://jevois.org/
• https://cloud.google.com/edge-tpu/
• https://www.nvidia.com/en-us/autonomous-machines/embedded-systems/
• https://www.96boards.org/product/ultra96/ai/
• https://www.tensorflow.org/lite/
• http://docs.openmv.io/
• http://mpqa.cs.pitt.edu/opinionfinder/opinionfinder_2/
• https://en.wikipedia.org/wiki/AI_winter
• https://en.wikipedia.org/wiki/Computer_chess
• https://en.wikipedia.org/wiki/Watson_(computer)
• https://cloud.google.com/products/ai/
• https://stockfishchess.org/
• https://link.springer.com/chapter/10.1007%2F978-3-540-72079-9_10
• https://www.technologyreview.com/2017/04/11/5113/the-dark-secret-at-the-heart-of-ai/
• https://prodi.gy/
• https://github.com/mnielsen/neural-networks-and-deep-learning
• https://towardsdatascience.com/what-the-hell-is-perceptron-626217814f53
• http://deeplearning.stanford.edu/wiki/index.php/UFLDL_Tutorial
• https://www.arundo.com/
• https://www.canvass.io/
• https://c3.ai/
• https://www.uptake.com/

Applied Research paper/ Publication

• Microsoft research

  • Home Automation in the Wild: Challenges and Opportunities

• IBM research

  • https://www.ibm.com/quantum-computing/

• Google Machine learning

• Google research

• Adaptive Machine Learning forCredit Card Fraud Detection(PhD thesis paper)

• Book

  • Theory: Quantum Computation and Quantum Information: 10th Anniversary Edition, Michael Nielson, Isaac L. Chuang
  • AI blueprints
  • AI by example
  • AI with Python
  • AI in finance