Publications

Blind Confusion of Classification Networks

A black box evaluation under common and structured image corruptions

This work presents a black box robustness evaluation of image classification networks under common and structured image corruptions. It analyzes 37 models across 15 corruption types and introduces Accuracy Confidence Divergence as a diagnostic measure for comparing accuracy degradation with confidence behavior.

Atam O. Erkara, Markus Mayer, Blind confusion of classification networks: A black box evaluation under common and structured image corruptions, Neurocomputing, Volume 686, 2026, 133678, ISSN 0925-2312.

https://doi.org/10.1016/j.neucom.2026.133678
Read Paper →View Code →

Projects

CryptoCore ECC Arithmetic cover

CryptoCore ECC Arithmetic

Embedded cryptography case study on the DE1-SoC platform using the CryptoCore hardware accelerator. The work focuses on GF(p)-based ECC arithmetic, Jacobian coordinates, Montgomery domain transformation, Point Doubling, Point Addition, Point Multiplication, Linux driver integration, and SageMath based verification.

Reinforcement Learning Reliability cover

Reinforcement Learning Reliability

Reliability study of continuous-control reinforcement learning in Lunar Lander. The work starts with DDPG under wind and turbulence, then shifts to PPO for faster stress testing under random force, low gravity, and high disturbance. Frame stacking and observation stacking are evaluated as practical robustness improvements.

HandsON Music App cover

HandsON Music App

Real-time hand gesture control for music playback using MediaPipe hand tracking. The app includes gesture-driven playback control and an in-page usage guide.

Full Stack Web Development cover

Full Stack Web Development

Full-stack Next.js project built with TypeScript, React, MongoDB Atlas, and Mongoose. It includes responsive UI, multilingual content handling, and role-based access for Guest, User, and Admin workflows.

FPGA Data Processing Pipeline cover

FPGA Data Processing Pipeline

VHDL project on the DE1-SoC board for deterministic data processing. Static financial values are read from ROM, stored in on-chip memory, processed with exponential smoothing through custom IEEE-754 based single-precision arithmetic modules, and displayed through the seven-segment interface.

CryptoCore ECC Arithmetic cover

CryptoCore ECC Arithmetic

Embedded cryptography case study on the DE1-SoC platform using the CryptoCore hardware accelerator. The work focuses on GF(p)-based ECC arithmetic, Jacobian coordinates, Montgomery domain transformation, Point Doubling, Point Addition, Point Multiplication, Linux driver integration, and SageMath based verification.

Reinforcement Learning Reliability cover

Reinforcement Learning Reliability

Reliability study of continuous-control reinforcement learning in Lunar Lander. The work starts with DDPG under wind and turbulence, then shifts to PPO for faster stress testing under random force, low gravity, and high disturbance. Frame stacking and observation stacking are evaluated as practical robustness improvements.

Research and Engineering

Support Vector Machines illustration

ML Concept Breakdown

Support Vector Machines

Visual explanation of SVM for classification problems. Covers margin based separation, support vectors, soft and hard margins, kernel trick, and a penguin dataset example.

Open Insight →GitHub Notebook →
SOME/IP Service Discovery Analysis illustration

Automotive Ethernet

SOME/IP Service Discovery Analysis

Case study on SOME/IP and SOME/IP-SD traffic analysis with Vector CANoe, VN5610, ANDi, Wireshark, and Technica MediaGateway. The work focuses on CAMF and ADAS service discovery, packet inspection, and partial reproduction of provider-side service behaviour.

Open Page →
TPM2 Secure Storage Study illustration

Embedded Security

TPM2 Secure Storage Study

TPM2 and Linux encryption study using tpm2-tools on Ubuntu with Hyper-V vTPM. Covers boot security concepts, PCR measurements, key sealing, persistent handles, RSA based file encryption, and an incomplete LUKS integration attempt.

Open Page →

Certificates

NVIDIA logo

Certificate

Building Real-Time Video AI Applications

NVIDIAIssued on November 15, 2025

Credential ID

Q4EMbY4-RIC3lBbVLPFPYQ
TAO ToolkitNVIDIA DeepStream SDKVideo AIComputer VisionReal-time inference