no code implementations • 11 May 2023 • Shakti N. Wadekar, Eugenio Culurciello
Preliminary results are shown on natural and medical image datasets.
1 code implementation • 26 May 2022 • Andre Xian Ming Chang, Parth Khopkar, Bashar Romanous, Abhishek Chaurasia, Patrick Estep, Skyler Windh, Doug Vanesko, Sheik Dawood Beer Mohideen, Eugenio Culurciello
In this work we propose a Reinforcement Learning framework with Global Graph Attention (GGA) module and output masking of invalid placements to find and optimize instruction schedules.
no code implementations • 8 Feb 2020 • Thomas Molnar, Eugenio Culurciello
Other approaches for navigating sparse environments require intrinsic reward generators, such as the Intrinsic Curiosity Module (ICM) and Augmented Curiosity Modules (ACM).
1 code implementation • 24 May 2019 • Vincenzo Lomonaco, Karan Desai, Eugenio Culurciello, Davide Maltoni
High-dimensional always-changing environments constitute a hard challenge for current reinforcement learning techniques.
1 code implementation • NeurIPS 2018 • Kuan Han, Haiguang Wen, Yizhen Zhang, Di Fu, Eugenio Culurciello, Zhongming Liu
When unfolded over time, the recurrent processing gives rise to an increasingly deeper hierarchy of non-linear transformation, allowing a shallow network to dynamically extend itself into an arbitrarily deep network.
no code implementations • ICML 2018 • Haiguang Wen, Kuan Han, Junxing Shi, Yizhen Zhang, Eugenio Culurciello, Zhongming Liu
Given image input, PCN runs recursive cycles of bottom-up and top-down computation to update its internal representations and reduce the difference between bottom-up input and top-down prediction at every layer.
no code implementations • 8 Aug 2017 • Vinayak Gokhale, Aliasger Zaidy, Andre Xian Ming Chang, Eugenio Culurciello
Snowflake is able to achieve a computational efficiency of over 91% on modern CNN models.
Hardware Architecture
no code implementations • 1 Aug 2017 • Andre Xian Ming Chang, Aliasger Zaidy, Vinayak Gokhale, Eugenio Culurciello
Given a programmable hardware accelerator with a CNN oriented custom instructions set, the compiler's task is to exploit the hardware's full potential, while abiding with the hardware constraints and maintaining generality to run different CNN models with varying workload properties.
14 code implementations • 14 Jun 2017 • Abhishek Chaurasia, Eugenio Culurciello
As a result they are huge in terms of parameters and number of operations; hence slow too.
no code implementations • 8 Jun 2017 • Alfredo Canziani, Eugenio Culurciello
In the past five years we have observed the rise of incredibly well performing feed-forward neural networks trained supervisedly for vision related tasks.
49 code implementations • 7 Jun 2016 • Adam Paszke, Abhishek Chaurasia, Sangpil Kim, Eugenio Culurciello
The ability to perform pixel-wise semantic segmentation in real-time is of paramount importance in mobile applications.
Ranked #10 on Semantic Segmentation on ScanNetV2
4 code implementations • 24 May 2016 • Alfredo Canziani, Adam Paszke, Eugenio Culurciello
Since the emergence of Deep Neural Networks (DNNs) as a prominent technique in the field of computer vision, the ImageNet classification challenge has played a major role in advancing the state-of-the-art.
no code implementations • 19 Nov 2015 • Jonghoon Jin, Aysegul Dundar, Eugenio Culurciello
Recent studies have shown that Convolutional Neural Networks (CNNs) are vulnerable to a small perturbation of input called "adversarial examples".
no code implementations • 19 Nov 2015 • Aysegul Dundar, Jonghoon Jin, Eugenio Culurciello
In this work, we propose to train a deep convolutional network based on an enhanced version of the k-means clustering algorithm, which reduces the number of correlated parameters in the form of similar filters, and thus increases test categorization accuracy.
Ranked #69 on Image Classification on MNIST
1 code implementation • 17 Nov 2015 • Andre Xian Ming Chang, Berin Martini, Eugenio Culurciello
Recurrent Neural Networks (RNNs) have the ability to retain memory and learn data sequences.
2 code implementations • 17 Dec 2014 • Jonghoon Jin, Aysegul Dundar, Eugenio Culurciello
We present flattened convolutional neural networks that are designed for fast feedforward execution.
1 code implementation • 1 Jun 2013 • Eugenio Culurciello, Jonghoon Jin, Aysegul Dundar, Jordan Bates
On the other hand, in unsupervised learning, one cannot rely on back-propagation techniques to learn the connections between layers.