AIMLSYSTEMS 2026

Detailed Technical Program

Research, Industry, Posters & Panel Presentations

Tue, Oct 6 Wed, Oct 7 Thu, Oct 8 Fri, Oct 9

Tuesday, October 6

Time (CET)Event DetailsLocation
18:00 - 18:30 Registration First Floor Lobby
18:30 - 19:30 Conference Opening & Welcome Reception
 First Floor Lobby

Wednesday, October 7

  
Time (CET)Session 1 / PlenarySession 2
08:30 - 09:00 Registration First Floor Lobby
09:00 - 10:00 Plenary Session: Prof. Michele Magno (ETH) Room A0.1
10:00 - 10:30 Coffee Break First Floor Lobby
10:30 - 12:30 Research Track 1
Room A0.1		 Industry Track 1
Room A0.2
10:30 - 10:50 LDT-Poly: Low-Degree Trainable Polynomial Activation for Privacy-Preserving Deep Learning
Alessandro Falcetta, Fabrizio Pittorino, Manuel Roveri		 Evaluating Low-Power Edge-AI for Always-on Contextual Vision
Francesca Palermo, Greta Corti, Aurelio Telli, Syed Shakib Sarwar, Derek Valleroy, Barbara De Salvo, Chiao Liu, Diana Trojaniello
10:50 - 11:10 Consensus Reservoirs for Synchronous Federated Optimization
Ojaswa Varshney, Pradeep Singh, Rajeev Shorey		 EdgeTalk: Agentic Speech-Text Models for Edge Devices
Megha Sharma, Tushar Vatsal, Prakhar Swarup, Raghavendra Bilgi
11:10 - 11:30 Memory Refinement for Latent Replays in Continual Learning
Edoardo Giudici, Fabrizio Pittorino, Manuel Roveri		 Bridging Photonics and AI: CVAE-Based Inverse Design of Plasmonic Sensors
Anish Datta, Soma Bandyopadhyay, Subhasri Chatterjee, Anirban Chaudhuri, Arpan Pal
11:30 - 11:50 S-VETT: Semantic VectorDB-Enabled Transfer Learning for Time-Series Forecasting Under Extreme Data Scarcity
Jonatan Sciaky, Lorenzo Epifani, Alessandro Falcetta, Manuel Roveri		 Text2Flowsheet: Generating DWSIM-Ready Flowsheets from Natural Language Descriptions
Amit Sangroya, Arushi Jain, Monika Sharma, Shubham Paliwal, Suparshva Jain
11:50 - 12:10 Malware families discovery via Open-Set Recognition on Android manifest permissions
Filippa Leveni, Matteo Mistura, Francesco Tubaldi, Carmine Giangregorio, Nicolò Pastore, Cesare Alippi, Giacomo Boracchi		 From Papers to Animations: A Plan Mediated LLM Pipeline with Parser Friendly Code Generation (PAGE)
Chitranshu Harbola, Balaji Srinivasan
12:10 - 12:30 DEMO: Live-Demo: RISC-V Edge Inference for Real-Time Eye-Movement Control on GAPses Smart Glasses Efficient Multimodal Document Retrieval in Industrial Settings via Compressed Vision-Language Embeddings
Andrea Sassella, Mark J. Carman
12:30 - 13:30 Lunch First Floor Lobby
13:30 - 14:30 Keynote Talk: Prof. Cesare Alippi Room A0.1
14:30 - 15:00 Sponsor Talk: Dr. Joel Goodman (ONR) Room A0.1
15:00 - 15:30 Coffee Break First Floor Lobby
15:30 - 16:50 Research Track 2
Room A0.1		 Industry Track 2
Room A0.2
15:30 - 15:50 YOSO-Rep ViT. Improving Real-Time Panoptic Segmentation with Rep ViT
Marc Ellenfeld, Maximilian Schwaiger, Moritz Ufer, Jonathan Kobold		 NL2SQL System Aided by Memory: Usage of Prior Reasonings for Faster inference
Harshita Chandwani, Venkata Rama Krishna Sreepada, Sai Karthik Athota
15:50 - 16:10 Dynamic Reweighting for Instance to Model-based Transition for On-Device Speaker Verification
Michele Craighero, Matteo Figini, Manuel Pariente, Giacomo Boracchi, Manuel Roveri		 Exploiting Cross-Query Redundancy in Multi-Source Workloads Using Hyper-DAGs and Adaptive Operator DAG
Ravi Kumar Singh, Shruti Kunde, Rekha Singhal
16:10 - 16:30 Bridging Symbolic Models and Simulation: LLM-Generated FMUs for HVAC Control
Javad Hosseinzadeh, Mohammad Esmaeili, Sascha Hammes, Thomas Ederer, Gernot Steindl, Philipp Zech		 GPU-RAR-GPU Retrieval Platform at LinkedIn
Dhritiman Das, Vishal Shah, Chirag Mehta, Pratik Desai, Chinmay Naik, Manika Agarwal, Ronak Kaoshik, Shirisha Singh, Meng Zhang
16:30 - 16:50 TRACE: Target-Responsive Actionable Causal Engine: A CausalGraph-Guided System for Financial KPI Planning
Soubhik Bhattacharya, Aniruddha Bardhan, Siddharth Kumar, Arkadeep Banerjee		 Breaking the Tie: Personalized Reranking for Relevance-Saturated Educational User-Generated Content Search
Anbang Wu
16:50 - 17:00 Break (No Coffee)
17:00 - 18:00 Panel: Ethical AI Room A0.1
18:00 - 19:30
Poster Session & Aperitif First Floor Lobby
Wednesday Posters:
  • LDT-Poly: Low-Degree Trainable Polynomial Activation
  • Consensus Reservoirs for Synchronous Federated Optimization
  • Memory Refinement for Latent Replays in Continual Learning
  • S-VETT: Semantic VectorDB-Enabled Transfer Learning
  • Malware families discovery via Open-Set Recognition
  • Evaluating Low-Power Edge-AI for Contextual Vision
  • EdgeTalk: Agentic Speech-Text Models for Edge Devices
  • Bridging Photonics and AI: CVAE-Based Inverse Design
  • Text2Flowsheet: Generating DWSIM-Ready Flowsheets
  • From Papers to Animations: A Plan Mediated LLM Pipeline
  • YOSO-Rep ViT. Improving Real-Time Panoptic Segmentation
  • Dynamic Reweighting for Instance to Model-based Transition
  • Bridging Symbolic Models and Simulation: LLM-Generated FMUs
  • TRACE: Target-Responsive Actionable Causal Engine
  • NL2SQL System Aided by Memory
  • Exploiting Cross-Query Redundancy in Multi-Source Workloads
  • Efficient Multimodal Document Retrieval in Industrial Settings
  • GPU-RAR-GPU Retrieval Platform at LinkedIn
  • EWMA-CDT for Adaptive Temporal Aggregation in Online QIS Filtering
  • Reinforced Active Learning for Change Point Detection
19:30 - 21:30 Bus Departs at 20:15
Conference Dinner TBA

Thursday, October 8

Time (CET)Session 1 / PlenarySession 2
08:30 - 09:00 Registration First Floor Lobby
09:00 - 10:00 Plenary Session: Prof. Supratik Mukhopadhyay Room A0.1
10:00 - 10:30 Coffee Break First Floor Lobby
10:30 - 12:30 Research Track 3
Room A0.1		 Research Track 4
Room A0.2
10:30 - 10:50 Exploring Redundancy and Shared Representations for Transformer Models Optimization
Enrico Simionato, Federico Toschi, Andrea Sassella, Nicolò Brunello, Vincenzo Scotti, Mark J. Carman		 Beyond the Model: The Critical Role of Data Filtering in Clinical Machine Learning
Noah Subedar, Colin Campbell, Wenjing Zhang, Dan Perri, Sarah Culgin, Andrew Hamilton-Wright
10:50 - 11:10 Paired Replacement Differential Cache Maintenance for Dynamic Sparse and Mixture-of-Experts Inference
Mohammad Tanzil Idrisi		 Consistent Relexicalization of Clinical Documents using Graph-Based Approach
Dipankar Das, Atri Mandal, Sandeep Singh, Tushar Shandhilya
11:10 - 11:30 Adapter Slots Warp-Aligned Multi-LoRA Serving via WAR-Gated Kernel Promotion
Shashank Tippanavar, Vivek Yadav		 Assessing Reliability of Large Language Models for Clinical Summarization Across Indian Languages
Kunwar Zaid, Vikas Kumar, Jyotsana Khatri, Amit Sangroya
11:30 - 11:50 Energy-Efficient On-Device RAG on a Mobile NPU System Design and Benchmark on Snapdragon X Elite
Zhiyuan Cheng, Longying Lai		 Visualizing Medical Classifier Decision Boundaries via Latent Space Traversal
Tejas T. Singh, Rishu Kumar, Shreya Yadav, Rajiv Misra, Supratik Mukhopadhyay
11:50 - 12:10 A Cross-Stack Characterisation of Inter-Agent Message Compression in Multi-Agent LLM Systems
Akash Mondal		 Robust Lightweight Deep Learning Models for Oral Cancer Screening
Siddhant Bharadwaj, Aakash Shedsale, Tejashree Subramanya, Mohd. Azfar, Praveen Birur, Debnath Pal, Shankararama Sharma
12:10 - 12:30 Invited Presentation: EmbBERT: Attention under 2 MB memory
Riccardo Bravin, Massimo Pavan, Hazem Hesham Yousef Shalby, Fabrizio Pittorino, Manuel Roveri		 Leveraging Vision Foundation Models with a Lightweight PyraFuse Decoder for Skin-Fabric Semantic Segmentation
Jamal Saeedi, Alessandro Giusti
12:30 - 13:30 Lunch First Floor Lobby
13:30 - 14:30 Keynote Talk: Dr. Rajeev Rastogi Room A0.1
14:30 - 15:00 Sponsor Talk: Daniele Ceccarelli (Cisco) Room A0.1
15:00 - 15:30 Coffee Break First Floor Lobby
15:30 - 16:50 Research Track 5
Room A0.1		 Industry Track 3
Room A0.2
15:30 - 15:50 ACID-Bench: Fault-Injection Contracts for State-Changing Tool Agents
Andrei Vince		 Artificial Intelligent Application for Optical networking: an overview
Maurizio Gazzola
15:50 - 16:10 ParEvalLayer: When Partial LLM-Agent Evaluations Support a Decision
Wei-Jung Huang, Bonan Shen		 Substitutes, Not Complements: Entity Features vs Fine-tuned Embeddings for Expense Allocation
Sourav Karmakar, Reeti Sarup, Deepak Gujraniya, Mithun Ghosh
16:10 - 16:30 Invited Presentation: Beyond Labels: Self-Supervised Perception for Autonomous Driving
Carlo Sgaravatti, Daniele Kopyshevskiy, Davide Secco, Giacomo Boracchi		 Dynamic Pricing with Segment and Inventory Awareness via Decoupled Conversion Modeling
Aakash Swami, Narendhar Gugulothu
16:30 - 16:50 Engineering AI-Ready Data: Advanced Methods for Dataset Preparation
Diego Carrera, Stefano Bertolasi, Diego Stucchi, Beatrice Rossi, Pasqualina Fragneto		 CABIN: A Conditional Anchor-Based Framework for Add-on Pricing and Bundling
Aakash Swami, Srividhya Sethuraman
16:50 - 17:00 Break (No Coffee)
17:00 - 18:00 Panel: AI Agents Room A0.1
18:00 - 19:30
Poster Session & Aperitif First Floor Lobby
Thursday Posters:
  • Exploring Redundancy and Shared Representations...
  • Paired Replacement Differential Cache Maintenance...
  • Adapter Slots Warp-Aligned Multi-LoRA Serving
  • Energy-Efficient On-Device RAG on a Mobile NPU
  • Inter-Agent Message Compression in Multi-Agent LLMs
  • Tool Choice Confusion: Causal Minimal Tool Filtering
  • The Critical Role of Data Filtering in Clinical ML
  • Consistent Relexicalization of Clinical Documents
  • Reliability of LLMs for Clinical Summarization (Indian Langs)
  • Visualizing Medical Classifier Decision Boundaries
  • Robust Lightweight Deep Learning for Oral Cancer Screening
  • Leveraging Vision Foundation Models for Skin-Fabric Segmentation
  • ACID-Bench: Fault-Injection Contracts for Agents
  • ParEvalLayer: When Partial LLM-Agent Evaluations Support Decision
  • AI Application for Optical networking: an overview
  • Breaking the Tie: Personalized Reranking for Edu Search
  • Substitutes, Not Complements: Entity Features vs Fine-tuned
  • Dynamic Pricing with Segment and Inventory Awareness
  • CABIN: A Conditional Anchor-Based Framework for Pricing
  • Tracking What Disappears: Object Re-Identification in Waste Sorting
  • EmbBERT: Attention under 2 MB memory
  • Change Detection in Routed Optical Networks via Forecasting Residual Monitoring
  • Vision-Based Geometrical Measurement of Wire Ropes Using the Radon Transform
19:30 - 19:45 Conference Closing Room A0.1

Friday, October 9

Time (CET)Session 1 / PlenarySession 2
08:30 - 09:00 Registration First Floor Lobby
09:00 - 10:00 Plenary Session: Prof. Alberto Maria Metelli Room A0.1
10:00 - 10:30 Coffee Break First Floor Lobby
10:30 - 12:30 Workshop on Reinforcement Learning and Game Theory
Room A0.1		 Workshop on Explainable AI for Astrophysics
Room A0.2
12:30 - 13:30 Lunch First Floor Lobby
13:30 - 15:00 Workshop on Biomedical Imaging and Computer Vision
Room A0.1		 Workshop on Explainable AI for Astrophysics
Room A0.2
15:00 - 15:30 Coffee Break First Floor Lobby
15:30 - 17:00 Workshop on Biomedical Imaging and Computer Vision
Room A0.1