David S Kung

US Patent:

6601223, Jul 29, 2003

Filed:

Sep 29, 2000

Appl. No.:

09/675634

Inventors:

Ruchir Puri - New Rochelle NY
David S. Kung - Chappaqua NY
Anthony Drumm - Rochester MN

Assignee:

International Business Machines Corporation - Armonk NY

International Classification:

G06F 1750

US Classification:

716 6, 716 1, 716 2, 716 4, 716 5

Abstract:

A system and method are proposed for estimating interconnect delay in an Integrated Circuit (IC). A formula for effective capacitance is derived which considers the effect of slew as well as resistive shielding of capacitance, thus yielding more accurate delays for both the interconnects and the source driver (transistor gate). In the system and method, a resistor-capacitor (RC) tree model is used for iterative calculations of effective capacitance and slew for each RC tree node. The effective capacitance is determined for each node by proceeding outward from the source to the sinks, and the slew for each node is determined, using the effective capacitances just determined, by proceeding inward from the sinks to the source node. Once the source node slew determined at a previous iteration is within a specified threshold of the source node slew in the present iteration, the method stops and stores the present iteration values as the final estimates.

US Patent:

20210248765, Aug 12, 2021

Filed:

Feb 7, 2020

Appl. No.:

16/785469

Inventors:

- Armonk NY, US
Conrad M. ALBRECHT - White Plains NY, US
Siyuan LU - Yorktown Heights NY, US
Wei ZHANG - Elmsford NY, US
Ulrich Alfons FINKLER - Mahopac NY, US
David S. KUNG - Chappaqua NY, US
Xiaodong CUI - Chappaqua NY, US
Marcus FREITAG - Pleasantville NY, US

International Classification:

G06T 7/55
G06N 20/20
G06N 3/08
G06T 7/33
G06N 3/04

Abstract:

Techniques for image processing and transformation are provided. A plurality of images and a plurality of maps are received, and a system of neural networks is trained based on the plurality of images and the plurality of maps. A first image is received, and a first map is generated by processing the first image using the system of neural networks.

US Patent:

20200226425, Jul 16, 2020

Filed:

Jan 11, 2019

Appl. No.:

16/245489

Inventors:

- Armonk NY, US
Shiyu Chang - Elmsford NY, US
Mo Yu - White Plains NY, US
David S. Kung - Chappaqua NY, US

International Classification:

G06K 9/62
G06N 3/08
G06K 9/03

Abstract:

Techniques regarding a zero-confidence adversarial attack are provided. For example, one or more embodiments described herein can comprise a system, which can comprise a memory that can store computer executable components. The system can also comprise a processor, operably coupled to the memory, and that can execute the computer executable components stored in the memory. The computer executable components can comprise an adversarial component that computes a perturbation that causes misclassification by a neural network classifier. The computer executable components can also comprise a restoration component that determines a normal vector to a constraint contour developed by the neural network classifier. Further, the computer executable components can comprise a projection component that determines a tangential vector to the constraint contour.