Forschungsseminar
Das Forschungsseminar richtet sich an interessierte Studierende des Master- oder Bachelorstudiums. Andere Interessenten sind jedoch jederzeit herzlich willkommen! Die vortragenden Studenten und Mitarbeiter der Professur KI stellen aktuelle forschungsorientierte Themen vor. Vorträge werden in der Regel in Englisch gehalten. Den genauen Termin einzelner Veranstaltungen entnehmen Sie bitte den Ankündigungen auf dieser Seite.
Informationen für Bachelor- und Masterstudenten
Die im Studium enthaltenen Seminarvorträge (das "Hauptseminar" im Studiengang Bachelor-IF/AIF bzw. das "Forschungsseminar" im Master) können im Rahmen dieser Veranstaltung durchgeführt werden. Beide Lehrveranstaltungen (Bachelor-Hauptseminar und Master-Forschungsseminar) haben das Ziel, dass die Teilnehmer selbstständig forschungsrelevantes Wissen erarbeiten und es anschließend im Rahmen eines Vortrages präsentieren. Von den Kandidaten wird ausreichendes Hintergrundwissen erwartet, das in der Regel durch die Teilnahme an den Vorlesungen Neurocomputing (ehem. Maschinelles Lernen) oder Neurokognition (I+II) erworben wird. Die Forschungsthemen stammen typischerweise aus den Bereichen Künstliche Intelligenz, Neurocomputing, Deep Reinforcement Learning, Neurokognition, Neurorobotische und intelligente Agenten in der virtuellen Realität. Andere Themenvorschläge sind aber ebenso herzlich willkommen!Das Seminar wird nach individueller Absprache durchgeführt. Interessierte Studenten können unverbindlich Prof. Hamker kontaktieren, wenn sie ein Interesse haben, bei uns eine der beiden Seminarveranstaltungen abzulegen.
Kommende Veranstaltungen
Bridging Conflicting Views on Eye Position Signals: A Neurocomputational Approach to Perisaccadic PerceptionNikolai Stocks Tue, 4. 11. 2025, 13:45, A10.208.1 and https://webroom.hrz.tu-chemnitz.de/gl/mic-cv7-ptw Saccades are an integral component of visual perception, yet the accuracy and role of eye position signals in the brain remain unclear. The classical model of perisaccadic perception posits that the dorsal visual system combines an imperfect eye position signal with visual input, leading to systematic perisaccadic mislocalizations under specific experimental conditions. However, neurophysiological studies of eye position information have produced seemingly conflicting results. One team of researchers observed the eye position signal directly in gain-field neurons in the lateral intraparietal area (LIP) and found them incompatible with the classical model. In contrast, another team reported evidence for an eye position signal consistent with the classical model, even showing that accurate eye position can be decoded from neural activity. We modeled two subpopulations of neurons in LIP receiving input from two different sources, one representing the corollary discharge containing predictive presaccadic signals, the other representing a slowly updating proprioceptive eye position signal. By decoding eye position from the neural activity of these subpopulations, we observed the model containing sufficient information to allow the decoder to accurately predict and track the perisaccadic eye position. Our findings reconcile the apparent contradiction between the different neurophysiological studies by providing a unified framework for understanding eye position signals in perisaccadic perception. Our results suggest that a combination of a late-updating proprioceptive signal and a predictive corollary discharge is sufficient for accurately decoding eye position. |
Vergangene Veranstaltungen
Interacting corticobasal ganglia-thalamocortical loops shape behavioral control through cognitive maps and shortcutsFred Hamker Tue, 28. 10. 2025, A10.208.1 and https://webroom.hrz.tu-chemnitz.de/gl/mic-cv7-ptw Control of behavior is often explained in terms of a dichotomy, with distinct neural circuits underlying goal-directed and habitual control, yet accumulating evidence suggests these processes are deeply intertwined. We propose a novel anatomically informed cognitive framework, motivated by interacting corticobasal ganglia-thalamocortical loops as observed in different mammals. The framework shifts the perspective from a strict dichotomy toward a continuous, integrated network where behavior emerges dynamically from interacting circuits. Decisions within each loop contribute contextual information, which is integrated with goal-related signals in the basal ganglia input, building a network of dependencies. Loop-bypassing shortcuts facilitate habit formation. Striatal integration hubs may function analogously to attention mechanisms in Transformer neural networks, a parallel we explore to clarify how a variety of behaviors can emerge from an integrated network. |
Evaluating Speculative Decoding in Large Language Models: A Comparative Study of MEDUSA and EAGLE Architectures on Hallucination RisksAlaa Alshroukh Tue, 21. 10. 2025, A10.208.1 and https://webroom.hrz.tu-chemnitz.de/gl/mic-cv7-ptw Large Language Models (LLMs) deliver strong performance on tasks such as summarization and question answering, but autoregressive decoding remains a latency bottleneck in real-time applications due to its sequential, compute-intensive nature. This bottleneck arises in part because each decoding step requires transferring the entire set of model parameters from high-bandwidth memory (HBM) to the accelerator's cache, which limits throughput and underutilizes computational resources. To address this challenge, two acceleration methods are examined: EAGLE (Extrapolation Algorithm for Greater Language-model Efficiency) - a speculative sampling framework, and MEDUSA - a multi-head inference approach. Both techniques reduce latency by generating multiple tokens per step. Performance is evaluated by comparing each accelerated model to its baseline on summarization tasks, enabling controlled within-model analyses of speed and output quality. Particular attention is given to hallucination - fluent yet factually unsupported output - since generating multiple tokens at once can amplify small inaccuracies into extended errors. Results indicate a clear speed - factuality trade-off: EAGLE achieves the lowest latency and highest throughput while broadly preserving source meaning, but exhibits reduced factual stability on dense inputs. Medusa is slower but demonstrates stronger adherence to the source and more consistent factuality. Overall, the findings emphasize that accelerated LLM decoding requires careful balancing of efficiency and reliability, especially in fact-sensitive applications. |
Part 2: Key Concepts from the Nengo Summer School 2025Erik Syniawa Tue, 14. 10. 2025, A10.208.1 This talk presents key concepts from the Nengo Summer School, focusing on the Neural Engineering Framework (NEF) and its implementation through Nengo. The NEF provides three core principles: representation (how populations encode information), computation (how connections compute functions), and dynamics (how recurrent networks implement differential equations). These principles enable the construction of large-scale spiking neural networks for cognitive modeling (SPAUN). The talk covers the Semantic Pointer Architecture (SPA), a cognitive architecture that uses compressed neural representations to connect low-level neural activity with high-level symbolic reasoning. It also examines Legendre Memory Units (LMUs), an efficient method for processing temporal information in neural networks. Additionally, I will present my project work: a biologically inspired architecture that combines a visual pathway with the basal ganglia to play different Atari games. |
Erweiterung eines Aufmerksamkeitsmodells um eine gelernte IT-SchichtLucas Berthold Thu, 11. 9. 2025, 1/336 and https://webroom.hrz.tu-chemnitz.de/gl/mic-cv7-ptw Es gehört noch zu den zentralen offenen Fragen der Neurowissenschaften, warum die Repräsentationen von Objekten im visuellen System des Menschen stabil bleiben, auch wenn sich die Augen ständig bewegen. Es wird angenommen, dass der inferotemporale Kortex (IT) eine entscheidende Rolle bei der Objekterkennung spielt, indem dort Objekte in Form von stabilen Repräsentationen kodiert werden (Li und DiCarlo, 2008). Trotz der Robustheit dieser Repräsentationen gegenüber Veränderungen der Position eines Objekts auf der Retina haben Li und DiCarlo (2008) gezeigt, dass durch gezielte Manipulation an einer bestimmten retinalen Position einzelne IT-Neuronen, die zuvor selektiv auf bestimmte Objekte reagierten, nach der Manipulation an Selektivität verloren haben. Die zentrale Frage dieser Arbeit ist, ob sich die von Li und DiCarlo (2008) beobachteten Effekte auch in einem künstlichen neuronalen Modell nachbilden lassen, wenn allein die Verbindungen eines höheren visuellen Areals von einem bestehenden Aufmerksamkeitsmodell (Beuth, 2019) zu einer IT-Schicht, unter Berücksichtigung von lateraler Inhibition, gelernt werden. Dazu ist erforderlich, dass die IT-Neuronen sowohl selektiv für bestimmte Objekte als auch positionsinvariant sind. Die Ergebnisse der Arbeit zeigen, dass die Neuronen der gelernten IT-Schicht objektspezifisch und positionsinvariant selektieren können. Die Beobachtungen aus dem Manipulationsexperiment von Li und DiCarlo (2008) konnten jedoch nicht repliziert werden. |
Part 1: Key Concepts from the Nengo Summer School 2025Erik Syniawa Thu, 17. 7. 2025, 1/336 and https://webroom.hrz.tu-chemnitz.de/gl/mic-cv7-ptw This talk presents key concepts from the Nengo Summer School, focusing on the Neural Engineering Framework (NEF) and its implementation through Nengo. The NEF provides three core principles: representation (how populations encode information), computation (how connections compute functions), and dynamics (how recurrent networks implement differential equations). These principles enable the construction of large-scale spiking neural networks for cognitive modeling (SPAUN). The talk covers the Semantic Pointer Architecture (SPA), a cognitive architecture that uses compressed neural representations to connect low-level neural activity with high-level symbolic reasoning. It also examines Legendre Memory Units (LMUs), an efficient method for processing temporal information in neural networks. Additionally, I will present my project work: a biologically inspired architecture that combines a visual pathway with the basal ganglia to play different Atari games. |
Influence of Tokenization Strategies on the Prediction of CAD Model DescriptionsSayeda Hadisa Habibi Thu, 3. 7. 2025, 1/336 and https://webroom.hrz.tu-chemnitz.de/gl/mic-cv7-ptw Computer-aided design (CAD) significantly impacts manufacturing, architecture, and engineering industries. Automating the generation and modification of 3D CAD models helps reduce repetitive tasks, saves resources, and allows engineers to focus on innovation and complex design challenges. Modern CAD tools like Fusion 360 support programmatic design modifications, enabling AI-based automation in design workflows. As CAD systems become more advanced, there is a growing for deep learning methods that understand and manipulate CAD data. However, applying deep learning to CAD remains a challenge, primarily due to the difficulty of finding effective representations of 3D models. Traditional formats like voxels and point clouds are memory-intensive and lose geometric detail. The B-Rep format, while precise, is too complex to feed directly into neural networks. Moreover, tokenization methods like WordPiece, originally developed for text, struggle with the structured, continuous nature of CAD data. This research investigates CAD-specific tokenization approaches by comparing the standard WordPiece method with the sketch-extrude-based technique introduced by DeepCAD, which converts CAD models into text-based representations of design steps, making them suitable for neural network architectures. A masked token prediction task using a TinyBERT-style model is used to evaluate both methods in the context of CAD understanding and representation. |
Parameter-Efficient Fine-tuning of LLMs for Domain-Specific Programming LanguagesSupriya Dravid Tue, 3. 6. 2025, https://meet.google.com/xvk-gnhi-vxh This thesis investigated how Large Language Models (LLMs) could be adapted to generate reliable code for domain-specific languages, with a focus on ABAP, the core programming language used in SAP systems. Unlike widely used general-purpose languages, domain-specific languages like ABAP are significantly underrepresented in public datasets. This lack of representation poses a challenge, as zero-shot code generation often results in outputs that are inaccurate, incomplete, or misaligned with enterprise coding standards. To address this issue, the study explored parameter-efficient fine-tuning techniques, LoRA (Low-Rank Adaptation) and QLoRA (Quantized LoRA). These methods selectively fine-tuned small subsets of parameters while keeping the original pretrained model weights mostly unchanged. This approach enabled effective domain adaptation with much lower computational and memory overhead compared to full-model retraining. The methods were applied to small and mid-sized open-source models such as Phi-3-mini and Codegen-2B, and their performance was evaluated using metrics including Pass@k, CodeBLEU, and perplexity. In addition to this, the research included qualitative assessments from experienced ABAP developers, who rated the generated code based on functional correctness, performance, robustness, and maintainability. The results demonstrated that fine-tuned models significantly outperformed zero-shot baselines across all evaluation criteria. This work provides practical insights into how LLMs could be adapted for specialized programming environments, offering a cost-effective pathway for integrating generative AI into enterprise software development workflows. |
