Technical Program

Click on the icon to view the manuscript on IEEE XPlore in the IEEE ICASSP 2020 Open Preview.

COLL-L2: Session 3R: Robustness Reproducibility Replicability

Session Type: Lecture
Time: Friday, 8 May, 08:00 - 10:00
Location: On-Demand
Virtual Session: View on Virtual Platform
Session Chairs: Tulay Adali, University of Maryland, Baltimore County and Peter Schreier, Universität Paderborn, Germany
 
 COLL-L2.1: THE EMPIRICAL DUALITY GAP OF CONSTRAINED STATISTICAL LEARNING
         Luiz Chamon; University of Pennsylvania
         Santiago Paternain; University of Pennsylvania
         Miguel Calvo-Fullana; University of Pennsylvania
         Alejandro Ribeiro; University of Pennsylvania
 
 COLL-L2.2: CONTEXT AND UNCERTAINTY MODELING FOR ONLINE SPEAKER CHANGE DETECTION
         Hagai Aronowitz; IBM Research AI
         Weizhong Zhu; IBM Research AI
 
 COLL-L2.3: MODELING UNCERTAINTY IN PREDICTING EMOTIONAL ATTRIBUTES FROM SPONTANEOUS SPEECH
         Kusha Sridhar; University of Texas at Austin
         Carlos Busso; University of Texas at Austin
 
 COLL-L2.4: ACCURACY-ROBUSTNESS TRADE-OFF FOR POSITIVELY WEIGHTED NEURAL NETWORKS
         Ana Neacsu; University Politehnica of Bucharest
         Jean-Christophe Pesquet; Université Paris-Saclay, CentraleSupélec, Inria
         Corneliu Burileanu; University Politehnica of Bucharest
 
 COLL-L2.5: TOWARDS A NEW UNDERSTANDING OF THE TRAINING OF NEURAL NETWORKS WITH MISLABELED TRAINING DATA
         Herbert Gish; Raytheon BBN Technologies
         Jan Silovsky; Apple
         Man-Ling Sung; Chinese University of Hong Kong
         Man-Hung Siu; Apple
         William Hartmann; Raytheon BBN Technologies
         Zhuolin Jiang; Raytheon BBN Technologies
 
 COLL-L2.6: ON NETWORK SCIENCE AND MUTUAL INFORMATION FOR EXPLAINING DEEP NEURAL NETWORKS
         Brian Davis; Carnegie Mellon University
         Umang Bhatt; Carnegie Mellon University
         Kartikeya Bhardwaj; Carnegie Mellon University
         Radu Marculescu; Carnegie Mellon University
         José Moura; Carnegie Mellon University