Johanes Latupapua

Johanes Latupapua

Pasar Minggu, Jakarta Raya, Indonesia
2 rb pengikut 500+ koneksi

Gabung untuk melihat profilnya

Tentang

A strong experiences data specialist, an IT supervisor who passionate in data science…

Aktivitas

Bergabung sekarang untuk melihat semua aktivitas

Pengalaman

  • Gambar AIForesee

    AIForesee

    Jakarta, Indonesia

  • -

    Jakarta, Indonesia

  • -

    Jakarta, Jakarta, Indonesia

  • -

    Jakarta, Indonesia

  • -

  • -

    Jakarta, Indonesia

  • -

    Greater Jakarta Area, Indonesia

  • -

    Greater Jakarta Area, Indonesia

  • -

    Greater Jakarta Area, Indonesia

Pendidikan

  • Gambar BINUS University

    BINUS UNIVERSITY GRADUATE PROGRAM

    -

    Research thesis at deep learning domain that evaluated performance of Convolutional Neural Networks and optimizers on wildlife animal classification.

Lisensi dan Sertifikasi

Pengalaman Sukarela

  • Gambar Wahana Visi Indonesia

    English Translator

    Wahana Visi Indonesia

    - 1 bulan

    Social Services

    English translator in one day Sponsor visit from HKG

  • Gambar Markoding

    Co Facilitator

    Markoding

    - 1 bulan

    Education

    One-day Co facilitator at Markoding Innovation Challenge

  • Gambar Dealls – Jobs & Mentoring (YC W22)

    Data Super Mentor

    Dealls – Jobs & Mentoring (YC W22)

    - 2 bulan

    Social Services

    Mentored one student that has a strong passionate to have a career in data analytics!

Penerbitan

  • Performance Evaluation of Convolutional Neural Networks and Its Optimizers on Wildlife Animal Classification

    International Journal of Advanced Trends in Computer Science and Engineering

    Selecting optimizer and hyperparamater settings are able to affect accuracy achieved significantly. Selected values are generally performed by trial and error. This study evaluates performance of Convolutional Neural Network (CNN) and two optimizers combined with changes in 6 learning rate values. Researched optimizers are Adaptive Moment Estimation and RMS Prop in wildlife animal classification domain. Other hyperparameter values are set same. The architectures are implemented on DenseNet 121,…

    Selecting optimizer and hyperparamater settings are able to affect accuracy achieved significantly. Selected values are generally performed by trial and error. This study evaluates performance of Convolutional Neural Network (CNN) and two optimizers combined with changes in 6 learning rate values. Researched optimizers are Adaptive Moment Estimation and RMS Prop in wildlife animal classification domain. Other hyperparameter values are set same. The architectures are implemented on DenseNet 121, ResNet 50 and AlexNet. This study recommends best value of learning rate for training process which will be conducted with conditions similar to this study and provide other insights during study. The images in this study used wild animals with various positions from front side, back side and some part of bodies that describe real condition. In contradiction to other studies which generally use images in the hundreds of thousands to millions, the number of images used is 47, 841 taken from the Serengeti Season 1 Snapshot with imbalance conditions. Obtained result with the highest recall reached 79%.

    Lihat penerbitan

  • Business Intelligence for Employment Classification in Jakarta Government Data

    2019 International Conference on ICT for Smart Society (ICISS)

    As an open data which are provided by government, public can access and assist government to analyze and look for solution to social problems among communities. Jakarta Open Data shares employment data that is able to be starting point into analysis or recommendation by building data warehouse. OLTP data suggested for the data warehouse is the existing employment type based transactions, which will be used to analyze employment and unemployment rate in each region. The proposed method to design…

    As an open data which are provided by government, public can access and assist government to analyze and look for solution to social problems among communities. Jakarta Open Data shares employment data that is able to be starting point into analysis or recommendation by building data warehouse. OLTP data suggested for the data warehouse is the existing employment type based transactions, which will be used to analyze employment and unemployment rate in each region. The proposed method to design the data warehouse is using 9-steps from Kimball. The selected process for the data warehouse is calculating number of employment type, with dimensions of region, education type, sex and employment type. Region has 3 level area in sub-district and village. The result of process and analysis will be provided into certain reports and dashboard. Dashboard will be very helpful directly to provide high level data for all stakeholder or decision maker.

    Penulis lainnya
    • Kristian Wahyudi
    • Ritchie Chandra
    • Abba Suganda Girsang
    Lihat penerbitan

Kursus

  • Dphi Deep Learning Foundation Bootcamp

    ba1581b4-ef02-4e5e-b6b4-4

Proyek

  • Collection Score

    -

    Objective of the project is to creating score for customer who has debt or arrears according to payment history information, follow up to customer activities. The model will help collection staff, collection head to monitor collection process. My role was assisting to cleansing raw data from various resources, identified data could having interconnection and relevancy.

  • Credit Risk Scoring Model

    -

    This model was created for customer scoring model of new customer when she/he proposed new motorcycle credit loan that was composed from 21 features of customer profiles on its initial version. Model has been scaled into production, officially used by credit analysts on April 2021. Later model grows up with 2-3 features added and separated into motorcycle and car loan.

  • Loan Prediction by comparing 5 algorithms

    -

    I tried to create simple loan prediction by comparing 5 algorithms with Kaggle dataset. There are 13 features with final decision is loan will be approved or not. Original idea was taken from Ajaymanwani's effort.

    Lihat proyek

  • Predict Banknotes whether authentic or not

    -

    Simple Classification using Random Forest, Logistic Regression and SVM This is simple classification to predict whether banknotes are authentic or not. It was part of Dphi DL Bootcamps assignment which used one of popular datasets, UCI Machine Learning Repository. Some ideas comes from learning at bootcamps and other open sources. By using this dataset, I compared result of Random Forest and SVM. Happy learning all and keep growing!

    Lihat proyek

  • Performance Evaluation of CNN and Optimizer on Wildlife Animal Classification

    -

    This is my research thesis to complete my education as Master of Computer Science at deep learning domain. The motivation of this study is to search best learning rate to achieve highest accuracy/recall in wildlife animal classification while using 3 state of the art CNN model and Adam/RMSProp optimizer.

    Lihat proyek

  • Business Intelligence for Employment Classification in Jakarta Government Data

    -

    This is team paper work based on our assignment at Data Warehouse class. I worked with Kristian and Ritchie, my colleagues using Pentaho to extract, transform, load and analyze one citizenship dataset from Jakarta Open Data (https://data.jakarta.go.id/dataset). Two supervisors, our lecturers then proposed to an International conference and approved to be published there.

    Kreator lainnya
    • Kristian Wahyudi
    • Ritchie Chandra
    • Abba Suganda Girsang
    • Sani Muhamad Isa
    Lihat proyek

Bahasa

  • English

    Tingkat profesional mahir

  • Indonesian

    Tingkat fasih atau penutur asli

Organisasi

  • Jakarta Machine Learning

    Member

    - Saat ini

    Jakarta Machine Learning (JML) is a Jakarta-based nonprofit community, dedicated to connect Indonesian talents in Machine Learning from all over the world. We also have a mission to develop talent's capabilities in Machine Learning through Workshops and Focus Group Discussion. Our registered members ranging from Data Scientist, Consultant, Data Analyst, Business Analytics, Software Engineer, IT Lecturers, IT Manager and University students.

Aktivitas lainnya oleh Johanes

Lihat profil lengkap Johanes

  • Melihat siapa yang sama-sama Anda kenal
  • Minta diperkenalkan
  • Hubungi langsung Johanes
Bergabung untuk melihat profil lengkap

Profil serupa lainnya

Orang lain yang bernama Johanes Latupapua

2 orang lainnya yang bernama Johanes Latupapua ada di LinkedIn

Lihat orang lain yang bernama Johanes Latupapua

Tambahkan keahlian baru dengan kursus ini

Lihat semua kursus