Patents by Inventor Song Han

Song Han has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).

  • Patent number: 9923201
    Abstract: Provided herein are nanostructures for lithium ion battery electrodes and methods of fabrication. In some embodiments, a nanostructure template coated with a silicon coating is provided. The silicon coating may include a non-conformal, more porous layer and a conformal, denser layer on the non-conformal, more porous layer. In some embodiments, two different deposition processes, e.g., a PECVD layer to deposit the non-conformal layer and a thermal CVD process to deposit the conformal layer, are used. Anodes including the nanostructures have longer cycle lifetimes than anodes made using either a PECVD or thermal CVD method alone.
    Type: Grant
    Filed: May 12, 2015
    Date of Patent: March 20, 2018
    Assignee: Amprius, Inc.
    Inventors: Weijie Wang, Zuqin Liu, Song Han, Jonathan Bornstein, Constantin Ionel Stefan
  • Publication number: 20180046895
    Abstract: The present invention proposes a highly parallel solution for implementing ANN by sharing both weights matrix of ANN and input activation vectors. It significantly reduces the memory access operations, the on-chip buffers. In addition, the present invention considers how to achieve a load balance among a plurality of on-chip processing units being operated in parallel. It also considers a balance between the I/O bandwidth and calculation capabilities of the processing units.
    Type: Application
    Filed: August 22, 2016
    Publication date: February 15, 2018
    Inventors: Dongliang XIE, Junlong KANG, Song HAN
  • Publication number: 20180046905
    Abstract: The technical disclosure relates to artificial neural network. In particular, the technical disclosure relates to how to implement efficient data access control in the neural network hardware acceleration system. Specifically, it proposes an overall design of a device that can process data receiving, bit-width transformation and data storing. By employing the technical disclosure, neural network hardware acceleration system can avoid the data access process becomes the bottleneck in neural network computation.
    Type: Application
    Filed: December 26, 2016
    Publication date: February 15, 2018
    Inventors: Yubin LI, Song HAN, Yi SHAN
  • Publication number: 20180046914
    Abstract: The present invention relates to artificial neural networks, for example, deep neural networks. In particular, the present invention relates to a compression method considering load balance for deep neural networks and the device thereof. More specifically, the present invention relates to how to compress dense neural networks into sparse neural networks in an efficient way so as to improve utilization of resources of the hardware platform.
    Type: Application
    Filed: December 26, 2016
    Publication date: February 15, 2018
    Inventors: Xin LI, Song HAN, Zhilin LU, Yi SHAN
  • Publication number: 20180046901
    Abstract: The present technical disclosure relates to artificial neural networks, e.g., gated recurrent unit (GRU). In particular, the present technical disclosure relates to how to implement a hardware accelerator for compressed GRU based on an embedded FPGA. Specifically, it proposes an overall design processing method of matrix decoding, matrix-vector multiplication, vector accumulation and activation function. In another aspect, the present technical disclosure proposes an overall hardware design to implement and accelerate the above process.
    Type: Application
    Filed: December 27, 2016
    Publication date: February 15, 2018
    Inventors: Dongliang XIE, Song HAN, Yi SHAN
  • Publication number: 20180046897
    Abstract: The present invention relates to recurrent neural network. In particular, the present invention relates to how to implement and accelerate a recurrent neural network based on an embedded FPGA. Specifically, it proposes an overall design processing method of matrix decoding, matrix-vector multiplication, vector accumulation and activation function. In another aspect, the present invention proposes an overall hardware design to implement and accelerate the above process.
    Type: Application
    Filed: December 26, 2016
    Publication date: February 15, 2018
    Inventors: Junlong KANG, Song HAN, Yi Shan
  • Publication number: 20180046915
    Abstract: The present invention relates to artificial neural networks, for example, deep neural networks. In particular, the present invention relates to a compression method for deep neural networks with proper use of mask and the device thereof. More specifically, the present invention relates to how to compress dense neural networks into sparse neural networks while maintaining or even improving the accuracy of the neural networks after compression.
    Type: Application
    Filed: December 26, 2016
    Publication date: February 15, 2018
    Inventors: Shijie SUN, Song HAN, Xin LI, Yi SHAN
  • Publication number: 20180046919
    Abstract: The present invention relates to artificial neural networks, for example, deep neural networks. In particular, the present invention relates to a multi-iteration compression method for deep neural networks and the device thereof.
    Type: Application
    Filed: December 26, 2016
    Publication date: February 15, 2018
    Inventors: Xin LI, Song HAN, Shijie SUN, Yi SHAN
  • Patent number: 9878059
    Abstract: An aqueous approach to synthesize capped SnS quantum dots (QDs) followed by optional capping molecule extension by attaching one or more extending molecules to the capping molecule via peptide bond formation at elevated temperature. The capped SnS QDs may have a capping molecule:Sn:S molar ratio of 16:3:1 to 16:12:1. A suspension of SnS QDs was heat-treated at 200° C. for 0.5-4 hrs. The obtained SnS QDs showed an NIR emission peak at 820-835 nm with an excitation wavelength at 690 nm. The as synthesized SnS QDs were found to have high positive zeta potential of ˜30 mV and thus were toxic to cells. By neutralizing the SnS QDs the cytotoxicity was reduced to an accepted level. The heat-treatment step can be obviated by adding a glycerol solution containing S2? anions and capping molecule to a glycerol solution of Sn2+ ions.
    Type: Grant
    Filed: April 7, 2016
    Date of Patent: January 30, 2018
    Assignee: DREXEL UNIVERSITY
    Inventors: Wei-Heng Shih, Wan Y. Shih, Song Han, Xiaomin Niu, Shi Fang
  • Publication number: 20170338464
    Abstract: Provided are novel electrode material composite structures containing high capacity active materials formed into porous base structures. The structures also include shells that encapsulate these porous base structures. During lithiation of the active material, the shell mechanically constrains the porous base structure. The shell allows lithium ions to pass through but prevents electrolyte solvents from interacting with the encapsulated active material. In certain embodiments, the shell contains carbon, while the porous base structure contains silicon. Although silicon tends to swell during lithiation, the porosity of the base structure and/or void spaces inside the shell helps to accommodate this additional volume within the shell without breaking it or substantially increasing the overall size of the composite structure.
    Type: Application
    Filed: May 30, 2017
    Publication date: November 23, 2017
    Inventors: Rainer J. Fasching, Zuqin Liu, Song Han, Ghyrn E. Loveness, Constantin I. Stefan
  • Patent number: 9798971
    Abstract: Provided us a visual cortical circuit apparatus comprising: a current mirror unit which uses a transistor as a current source to generate a current having the same size as that of a reaction; a transconductance unit which takes, as an input, the current generated by the current mirror unit and outputs a voltage using a transconductance; and a buffer unit for converting the voltage output from the transconductance unit into a current and buffering the current.
    Type: Grant
    Filed: February 3, 2016
    Date of Patent: October 24, 2017
    Assignee: KOREA ADVANCED INSTITUTE OF SCIENCE AND TECHNOLOGY
    Inventors: Il Song Han, Woo Joon Han
  • Patent number: 9780365
    Abstract: Provided are battery electrode structures that maintain high mass loadings (i.e., large amounts per unit area) of high capacity active materials in the electrodes without deteriorating their cycling performance. These mass loading levels correspond to capacities per electrode unit area that are suitable for commercial electrodes even though the active materials are kept thin and generally below their fracture limits. A battery electrode structure may include multiple template layers. An initial template layer may include nanostructures attached to a substrate and have a controlled density. This initial layer may be formed using a controlled thickness source material layer provided, for example, on a substantially inert substrate. Additional one or more template layers are then formed over the initial layer resulting in a multilayer template structure with specific characteristics, such as a surface area, thickness, and porosity.
    Type: Grant
    Filed: October 20, 2011
    Date of Patent: October 3, 2017
    Assignee: Amprius, Inc.
    Inventors: Zuqin Liu, Song Han, Ghyrn E. Loveness
  • Publication number: 20170235298
    Abstract: A big data network or system for a process control system or plant includes a data storage device configured to receive process control data from control system devices and store the process control data. The big data network or system identifies various parameters or attributes from the process control data, and creates and uses rowkeys to store the parameters according to various combinations, such as combinations using timestamps. The big data network or system may also store certain aggregate data analyses associated with time periods specified by the timestamps. Accordingly, the big data network or system efficiently stores real-time data having measurements within a database schema, and users or administrators can leverage the aggregate data to analyze certain data associated with certain time periods.
    Type: Application
    Filed: April 28, 2017
    Publication date: August 17, 2017
    Inventors: Mark J. Nixon, Paul Richard Muston, Deji Chen, Song Han
  • Patent number: 9698410
    Abstract: Provided are novel electrode material composite structures containing high capacity active materials formed into porous base structures. The structures also include shells that encapsulate these porous base structures. During lithiation of the active material, the shell mechanically constrains the porous base structure. The shell allows lithium ions to pass through but prevents electrolyte solvents from interacting with the encapsulated active material. In certain embodiments, the shell contains carbon, while the porous base structure contains silicon. Although silicon tends to swell during lithiation, the porosity of the base structure and/or void spaces inside the shell helps to accommodate this additional volume within the shell without breaking it or substantially increasing the overall size of the composite structure.
    Type: Grant
    Filed: October 30, 2015
    Date of Patent: July 4, 2017
    Assignee: Amprius, Inc.
    Inventors: Rainer J. Fasching, Zuqin Liu, Song Han, Ghyrn E. Loveness, Constantin I. Stefan
  • Patent number: 9692056
    Abstract: Batteries having improved current collection are provided. In some implementations, an electrode structure of a battery may include an active material and two or more current collectors in electrical communication with the active material. In some implementations, an electrode structure of a battery may include two or more current collector layers. According to various implementations, the electrode structure may or may not include a current collector substrate. In some implementations, a battery anode includes a current collector substrate in electronic contact with nanostructured active material. In order to ensure that electronic communication between the active material and the current collector substrate is maintained throughout the life of the battery, a second electronically conductive path is provided in the form of a current collector layer over the nanostructured active material. The additional layer is thin and electronically conductive, and does not interfere adversely with battery operation.
    Type: Grant
    Filed: April 15, 2013
    Date of Patent: June 27, 2017
    Assignee: Amprius, Inc.
    Inventors: Zuqin Liu, Song Han
  • Patent number: 9665088
    Abstract: A big data network or system for a process control system or plant includes a data storage device configured to receive process control data from control system devices and store the process control data. The big data network or system identifies various parameters or attributes from the process control data, and creates and uses rowkeys to store the parameters according to various combinations, such as combinations using timestamps. The big data network or system may also store certain aggregate data analyses associated with time periods specified by the timestamps. Accordingly, the big data network or system efficiently stores real-time data having measurements within a database schema, and users or administrators can leverage the aggregate data to analyze certain data associated with certain time periods.
    Type: Grant
    Filed: January 31, 2014
    Date of Patent: May 30, 2017
    Assignee: FISHER-ROSEMOUNT SYSTEMS, INC.
    Inventors: Mark J. Nixon, Paul Richard Muston, Deji Chen, Song Han
  • Publication number: 20170098819
    Abstract: Provided herein are novel template electrode materials and structures for lithium ion cells. Related methods are also provided. According to various embodiments, an electrode can include a nanostructured template, an electrochemically active material layer coating the template, and a first intermediate layer between the nanostructured template and the electrochemically active material layer. In one arrangement, the nanostructured template includes silicide nanowires. The electrochemically active material may be any of silicon, tin, germanium, carbon, metal hydrides, silicides, phosphides, and nitrides. The first intermediate layer may facilitate adhesion between the nanostructured template and the electrochemically active material layer, electronic conductivity within the electrode, and/or stress relaxation between the nanostructured template and the electrochemically active material layer.
    Type: Application
    Filed: July 2, 2012
    Publication date: April 6, 2017
    Applicant: AMPRIUS INC.
    Inventors: Ghyrn E. Loveness, Song Han, Zuqin Liu
  • Publication number: 20160376318
    Abstract: A polypeptide, a DNA molecule encoding the polypeptide, a vector, a preparation method and a use therefor are disclosed. The polypeptide comprises an amino acid sequence represented by formula (I) or formula (II): formula (I) comprises an amino acid sequence represented by SEQ ID NO: 1; formula (II) comprises an amino acid sequence obtained by subjecting the amino acid sequence represented by SEQ ID NO: 1 to modification, substituted, and deletion or addition of one or more amino acids. The polypeptide may be used in the preparation of drugs that treat or prevent diseases related to infections caused by bacteria, and can also be used in the preparation of drugs that promote tissue repair and wound healing.
    Type: Application
    Filed: July 16, 2014
    Publication date: December 29, 2016
    Inventors: Wenxin LI, Zhijian CAO, Yingliang WU, Zesheng WANG, Song HAN
  • Publication number: 20160297839
    Abstract: An aqueous approach to synthesize capped SnS quantum dots (QDs) followed by optional capping molecule extension by attaching one or more extending molecules to the capping molecule via peptide bond formation at elevated temperature. The capped SnS QDs may have a capping molecule:Sn:S molar ratio of 16:3:1 to 16:12:1. A suspension of SnS QDs was heat-treated at 200° C. for 0.5-4 hrs. The obtained SnS QDs showed an NIR emission peak at 820-835 nm with an excitation wavelength at 690 nm. The as synthesized SnS QDs were found to have high positive zeta potential of ˜30 mV and thus were toxic to cells. By neutralizing the SnS QDs the cytotoxicity was reduced to an accepted level. The heat-treatment step can be obviated by adding a glycerol solution containing S2? anions and capping molecule to a glycerol solution of Sn2+ ions.
    Type: Application
    Filed: April 7, 2016
    Publication date: October 13, 2016
    Applicant: Drexel University
    Inventors: Wei-Heng Shih, Wan Y. Shih, Song Han, Xiaomin Niu, Shi Fang
  • Publication number: 20160190600
    Abstract: Provided are electrode layers for use in rechargeable batteries, such as lithium ion batteries, and related fabrication techniques. These electrode layers have interconnected hollow nanostructures that contain high capacity electrochemically active materials, such as silicon, tin, and germanium. In certain embodiments, a fabrication technique involves forming a nanoscale coating around multiple template structures and at least partially removing and/or shrinking these structures to form hollow cavities. These cavities provide space for the active materials of the nanostructures to swell into during battery cycling. This design helps to reduce the risk of pulverization and to maintain electrical contacts among the nanostructures. It also provides a very high surface area available ionic communication with the electrolyte. The nanostructures have nanoscale shells but may be substantially larger in other dimensions.
    Type: Application
    Filed: November 25, 2015
    Publication date: June 30, 2016
    Inventors: Yi Cui, Song Han, Ghyrn E. Loveness