لطفا منتظر بمانید ...
0% Complete
صفحه اصلی
/
سی و چهارمین کنفرانس بین المللی مهندسی برق
SPNP: A Calibration-Aware and Sparse Output Normalization for Deep Neural Networks
نویسندگان :
Morteza Taheri
1
1- دانشگاه زنجان
کلمات کلیدی :
SPNP،softmax replacement،probability calibration
چکیده :
In this paper, we introduce Shifted Power Normalized Probabilities (SPNP), a simple closed-form output activation function designed as an alternative to Softmax for probability normalization. The proposed method performs sample-wise standardization of logits, followed by a minimal shift and power-based normalization. SPNP improves probability calibration and enables controllable sparsity without modifying network architectures or the underlying loss formulation. Experiments on CIFAR-100, CIFAR-10, and MNIST, where all models are trained using the same standard cross-entropy loss applied to logits, demonstrate that SPNP significantly improves calibration metrics such as ECE, MCE, and the Brier score, while maintaining or improving classification accuracy. In addition, SPNP produces sparse output distributions and introduces negligible inference-time computational overhead, making it a practical and reliable alternative to Softmax for calibrated probabilistic inference.
لیست مقالات
لیست مقالات بایگانی شده
Optimal Bidding Strategy of a Cascade Hydroelectric Unit in a Day-ahead Energy Market Using Particle Swarm Optimization
Shabnam Ahmadian Titkanloo - Sahar Ahmadian Titkanloo - Asghar Akbari Foroud - Sadaf Ahmadian - Soheil Ahmadian Titkanloo
A Novel Supervisor Framework Toward Optimizing User Quality of Experience for Video Streaming in Software-Defined Networks
Nafiseh Fareghzadeh - Mahdi Bazargani - Sasan H. Alizadeh - Hasan Ziafat
Net Load Forecasting of Household Prosumers Considering Deep Reinforcement Learning
Behzad Motallebi Azar - Rasool Kazemzadeh - Morteza Zare Oskouei - Behnam Mohammadi-Ivatloo
Fabrication of reduced graphene oxide/Chitosan for detection of Amlodipine by Surface Plasmon Resonance
Hassan Nasiri - Erfan Hashemzadeh - Nima Mohammadzadeh Asl - Seyed Mohammad Reza Sadat Hosseini
Finite-Time Bipartite Time-Varying Formation tracking for Heterogeneous Nonlinear Multi-Agent Systems
Mohammad Reza Mehrabi Koushki - Javad Askari - Marzieh Kamali
New Single Phase Direct AC-AC Converters As A Series Static Digital Voltage Stabilizer With The Introduction Of Transformer And Transformerless Network Connection
Seyed mohsen Mortazavi - Reza Beiranvand
طراحی کنترل‌کننده استروب‌اسکوپ زمان واقعی مبتنی بر هوش مصنوعی برای سیستم های دورانی
مهدی مظفری - سعید جعفری نسب - حامد پورکاوه - سعید شمقدری
Image Inpainting Using AutoEncoder and Guided Selection of Predicted Pixels
Mohammad Hossein Givkashi - Mahshid Hadipour - َArezoo PariZanganeh - Zahra Nabizadeh Shahre-Babak - Nader Karimi - Shadrokh Samavi
Modeling of dielectrophoretic single-stage continuous separation of Escherichia coli K38 in a microfluidic channel
Saeed Saedy - Navid Alaei Sheini - Shahrzad Ajabi
الگوریتم تشخیصی برای طبقه بندی سرطان خون لوسمی لنفوسیتی حاد با استفاده از شبکه های عصبی عمیق در یادگیری آنلاین
رضا گودرزی - علی جلالی - امید هاشمی پورتفرشی
بیشتر
ثمین همایش، سامانه مدیریت کنفرانس ها و جشنواره ها - نگارش 44.7.2
