H. Hoffmann
A5103394599- Particle physics theoretical and experimental studies
- High-Energy Particle Collisions Research
- Quantum Chromodynamics and Particle Interactions
- Particle Detector Development and Performance
- Lung Cancer Diagnosis and Treatment
- Lung Cancer Treatments and Mutations
- Tracheal and airway disorders
- Medical Imaging and Pathology Studies
- Pleural and Pulmonary Diseases
- Cancer Diagnosis and Treatment
- Dark Matter and Cosmic Phenomena
- Lung Cancer Research Studies
- Robot Manipulation and Learning
- Metallurgy and Material Forming
- Trauma Management and Diagnosis
- Neural Networks and Applications
- Domain Adaptation and Few-Shot Learning
- Pregnancy and preeclampsia studies
- Nuclear reactor physics and engineering
- Congenital Diaphragmatic Hernia Studies
- Neuroblastoma Research and Treatments
- Neural dynamics and brain function
- Cancer Immunotherapy and Biomarkers
- Congenital Heart Disease Studies
- Nuclear Materials and Properties
Technical University of Munich
2019-2024
Klinikum rechts der Isar
2018-2024
University Medical Center
2024
University Hospital and Clinics
2024
Microwave Monolithics (United States)
2023
HRL Laboratories (United States)
2010-2022
TU Dresden
2003-2022
Helmholtz-Zentrum Dresden-Rossendorf
2021-2022
University of Siegen
2022
European Organization for Nuclear Research
1985-2020
Nonlinear dynamical systems have been used in many disciplines to model complex behaviors, including biological motor control, robotics, perception, economics, traffic prediction, and neuroscience. While often the unexpected emergent behavior of nonlinear is focus investigations, it equal importance create goal-directed (e.g., stable locomotion from a system coupled oscillators under perceptual guidance). Modeling with is, however, rather difficult due parameter sensitivity these systems,...
We provide a general approach for learning robotic motor skills from human demonstration. To represent an observed movement, non-linear differential equation is learned such that it reproduces this movement. Based on representation, we build library of movements by labeling each recorded movement according to task and context (e.g., grasping, placing, releasing). Our formulated generalization can be achieved simply adapting start goal parameter in the desired position values For object...
With the growing use of graph convolutional neural networks (GCNNs) comes need for explainability. In this paper, we introduce explainability methods GCNNs. We develop analogues three prominent networks: contrastive gradient-based (CG) saliency maps, Class Activation Mapping (CAM), and Excitation Back-Propagation (EB) their variants, gradient-weighted CAM (Grad-CAM) EB (c-EB). show a proof-of-concept these on classification problems in two application domains: visual scene graphs molecular...
The unprecedented success of deep neural networks in many applications has made these a prime target for adversarial exploitation. In this paper, we introduce benchmark technique detecting backdoor attacks (aka Trojan attacks) on convolutional (CNNs). We the concept Universal Litmus Patterns (ULPs), which enable one to reveal by feeding universal patterns network and analyzing output (i.e., classifying as `clean' or `corrupted'). This detection is fast because it requires only few forward...
Dynamical systems can generate movement trajectories that are robust against perturbations. This article presents an improved modification of the original dynamic primitive (DMP) framework by Ijspeert et al [1], [2]. The new equations generalize movements to targets without singularities and large accelerations. Furthermore, represent a in 3D task space depending on choice coordinate system (invariance under invertible affine transformations). Our modified DMP is motivated from biological...
Robots in a human environment need to be compliant. This compliance requires that preplanned movement can adapted an obstacle may moving or appearing unexpectedly. Here, we present general framework for generation and mid-flight adaptation obstacles. For robust motion generation, Ijspeert et al developed the of dynamic primitives [1], [2], [3], [4], which represent demonstrated with set differential equations. These equations allow adding perturbing force without sacrificing stability...
Uncontrolled proliferation is a hallmark of malignant tumour growth. Its prognostic role in non-small cell lung cancer (NSCLC) has been investigated numerous studies with controversial results. We aimed to resolve these controversies by assessing the Ki-67 index (PI) three large, independent NSCLC cohorts. Proliferation was retrospectively analysed immunohistochemistry cohort 1065 and correlated clinicopathological data including outcome therapy. Results were validated two cohorts 233...
Numerous studies have been published on single aspects of pulmonary adenocarcinoma (ADC). To comprehensively link clinically relevant ADC characteristics, we evaluated established morphological, diagnostic and predictive biomarkers in 425 resected ADCs. Morphology was reclassified. Cytokeratin-7, thyroid transcription factor (TTF)1, napsin A, thymidylate synthase excision repair cross-complementing rodent deficiency complementation group-1 expression, anaplastic lymphoma kinase...
Gaussian mixture models (GMM) are powerful parametric tools with many applications in machine learning and computer vision. Expectation maximization (EM) is the most popular algorithm for estimating GMM parameters. However, EM guarantees only convergence to a stationary point of log-likelihood function, which could be arbitrarily worse than optimal solution. Inspired by relationship between negative function Kullback-Leibler (KL) divergence, we propose an alternative formulation parameters...
Computational models of the neuromuscular system hold potential to allow us reach a deeper understanding function and clinical rehabilitation by complementing experimentation. By serving as means distill explore specific hypotheses, computational emerge from prior experimental data motivate future work. Here we review tools used understand including musculoskeletal modeling, machine learning, control theory, statistical model analysis. We conclude that these tools, when in combination, have...
Abstract Self-organized criticality (SOC) is a phenomenon observed in certain complex systems of multiple interacting components, e.g., neural networks, forest fires, and power grids, that produce power-law distributed avalanche sizes. Here, we report the surprising result avalanches from an SOC process can be used to solve non-convex optimization problems. To generate avalanches, use Abelian sandpile model on graph mirrors problem. For optimization, map areas onto search patterns for while...