- Number
- 20190393903
- Published
- 2019-12-26
- Filed
- 2018-06-29
- Assignee
- DISNEY ENTERPRISES, INC.
- Inventors
- MANDT; Stephan Marcel et al.
- CPC
- H03M7/30; G06N3/0499; G06N3/0475; G06N3/047; G06N3/0455; G06N3/045; G10L19/04; H03M1/00; G06N3/09; G06N3/08; H03M13/6502; H03M7/00
- Verdict
- Set aside generic ML compression (VAE encoding/decoding)
- Source
- Google Patents · FreePatentsOnline
Abstract
Embodiments include applying neural network technologies to encoding/decoding technologies by training and encoder model and a decoder model using a neural network. Neural network training is used to tune a neural network parameter for the encoder model and a neural network parameter for the decoder model that approximates an objective function. The common objective function may specify a minimized reconstruction error to be achieved by the encoder model and the decoder model when reconstructing (encoding then decoding) training data. The common objective function also specifies for the encoder and decoder models, a variable f representing static aspects of the training data and a set of variables z1:T representing dynamic aspects of the training data. During runtime, the trained encoder and decoder models are implemented by encoder and decoder machines to encode and decoder runtime sequences having a higher compression rate and a lower reconstruction error than in prior approaches.
Background
BACKGROUNDField of the Various Embodiments
Embodiments of the present invention generally relate to computer science and, more specifically, to efficient encoding and decoding sequences using variational autoencoders.DESCRIPTION OF THE RELATED ART
Codecs (COmpressor-DECompressor) are compression algorithms designed to encode/compress and decode/decompress source data (input sequences) to reduce the size of the source data. Many popular codecs produce lossy compression that reduces the data size of the source data by removing unnecessary or less important information of the source data. Such lossy codecs encode source data using inexact approximations and the partial discarding of the source data, but allows for a large reduction in data size when compared to uncompressed data or lossless compression. Lossy compression is commonly used to encode multimedia data sequences, such as audio or video, in a wide range of applications, such as playback from local storage and remote streaming over a network. Compression is highly useful as it reduces the data size of the source data, which significantly reduces the storage resources required to store the source data. Also, if the source data is to be transmitted over a network, compression allows for faster transmission of the source data and/or reduces the bit-rate and bandwidth requirements for transmitting the source data. A decoder algorithm is then used to decode/decompress the encoded/compressed data to reverse the en
Claims
1. A computer-implemented method for decoding an encoded sequence, the method comprising: receiving the encoded sequence at a decoder engine; and decoding the encoded sequence via a decoder model implemented by the decoder engine to generate a decoded sequence, wherein the decoder model is trained through neural network training to decode static aspects of the encoded sequence that are substantially consistent throughout the entirety of the encoded sequence.
2. The computer-implemented method of claim 1, wherein the encoded sequence comprises a plurality of ordered segments.
3. The computer-implemented method of claim 2, wherein the decoder model is further trained to decode dynamic aspects of a first segment included in the plurality of ordered segments that are unique to the first segment.
4. The computer-implemented method of claim 3, wherein the static aspects of the encoded sequence are decoded separately from the dynamic aspects of the first segment.
5. The computer-implemented method of claim 2, wherein the decoder model is further trained to decode dynamic aspects for each particular segment included in the plurality of ordered segments that are unique to the particular segment.
6. The computer-implemented method of claim 1, wherein the decoder model is generated through neural network training based on an objective function that reduces reconstruction error for encoding and decoding training input data.
7. The computer-implemented method of claim 1, wherein the decoder model is trained based on a decoder equation that specifies separate variables for decoding static aspects and dynamic aspects of encoded data.
8. The computer-implemented method of claim 1, wherein the decoder model is trained based on a decoder equation that expresses a Gaussian distribution.
9. The computer-implemented method of claim 1, further comprising receiving the encoded sequence from an encoder engine via a network, the encoder engine implementing an encoder model generated through neural network training.
10. A non-transitory computer-readable medium storing program instructions that, when executed by a processor, cause the processor to decode an encoded sequence by performing the steps of: receiving the encoded sequence at a decoder engine; and decoding the encoded sequence via a decoder model implemented by the decoder engine to generate a decoded sequence, wherein the decoder model is trained through neural network training to decode static aspects of the encoded sequence that are substantially consistent throughout the entirety of the encoded sequence.
11. The non-transitory computer-readable medium of claim 10, wherein the encoded sequence comprises a plurality of ordered segments.
12. The non-transitory computer-readable medium of claim 10, wherein the encoded sequence comprises a plurality of encoded video frames.
13. The non-transitory computer-readable medium of claim 10, wherein the encoded sequence comprises a plurality of encoded audio samples.
14. The non-transitory computer-readable medium of claim 11, wherein the decoder model is further trained to decode dynamic aspects of a first segment included in the plurality of ordered segments that are unique to the first segment.
15. The non-transitory computer-readable medium of claim 14, wherein the static aspects of the encoded sequence are decoded separately from the dynamic aspects of the first segment.
16. The non-transitory computer-readable medium of claim 10, wherein the decoder model is trained based on a decoder equation that expresses a Gaussian distribution.
17. The non-transitory computer-readable medium of claim 10, wherein the decoder model is trained based on a decoder equation that specifies separate variables for decoding static aspects and dynamic aspects of encoded data.
18. The non-transitory computer-readable medium of claim 10, further comprising receiving the encoded sequence from an encoder engine via a network, the encoder engine implementing an encoder model generated through neural network training.
19. A computing system configured for decoding an encoded sequence, comprising: a memory that includes a decoder engine; and a processor that is coupled to the memory and, upon executing the decoder engine, performs the steps of: receiving the encoded sequence; and decoding the encoded sequence via a decoder model implemented by the decoder engine to generate a decoded sequence, wherein the decoder model is trained through neural network training to decode static aspects of the encoded sequence that are substantially consistent throughout the entirety of the encoded sequence.
20. The computing system of claim 19, wherein the encoded sequence comprises a plurality of ordered segments.