Matches in SemOpenAlex for { <https://semopenalex.org/work/W3186789842> ?p ?o ?g. }
- W3186789842 abstract "The current standard for a variety of computer vision tasks using smaller numbers of labelled training examples is to fine-tune from weights pre-trained on a large image classification dataset such as ImageNet. The application of transfer learning and transfer learning methods tends to be rigidly binary. A model is either pre-trained or not pre-trained. Pre-training a model either increases performance or decreases it, the latter being defined as negative transfer. Application of L2-SP regularisation that decays the weights towards their pre-trained values is either applied or all weights are decayed towards 0. This paper re-examines these assumptions. Our recommendations are based on extensive empirical evaluation that demonstrate the application of a non-binary approach to achieve optimal results. (1) Achieving best performance on each individual dataset requires careful adjustment of various transfer learning hyperparameters not usually considered, including number of layers to transfer, different learning rates for different layers and different combinations of L2SP and L2 regularization. (2) Best practice can be achieved using a number of measures of how well the pre-trained weights fit the target dataset to guide optimal hyperparameters. We present methods for non-binary transfer learning including combining L2SP and L2 regularization and performing non-traditional fine-tuning hyperparameter searches. Finally we suggest heuristics for determining the optimal transfer learning hyperparameters. The benefits of using a non-binary approach are supported by final results that come close to or exceed state of the art performance on a variety of tasks that have traditionally been more difficult for transfer learning." @default.
- W3186789842 created "2021-08-02" @default.
- W3186789842 creator A5030379402 @default.
- W3186789842 creator A5063135412 @default.
- W3186789842 creator A5063366658 @default.
- W3186789842 date "2021-07-18" @default.
- W3186789842 modified "2023-09-27" @default.
- W3186789842 title "Non-binary deep transfer learning for image classification" @default.
- W3186789842 cites W1846799578 @default.
- W3186789842 cites W2047643928 @default.
- W3186789842 cites W2102605133 @default.
- W3186789842 cites W2104648049 @default.
- W3186789842 cites W2135346934 @default.
- W3186789842 cites W2138011018 @default.
- W3186789842 cites W2143668817 @default.
- W3186789842 cites W2149933564 @default.
- W3186789842 cites W2160921898 @default.
- W3186789842 cites W2194775991 @default.
- W3186789842 cites W2203224402 @default.
- W3186789842 cites W2253728219 @default.
- W3186789842 cites W2575615142 @default.
- W3186789842 cites W2591924527 @default.
- W3186789842 cites W2772989637 @default.
- W3186789842 cites W2798381792 @default.
- W3186789842 cites W2799041689 @default.
- W3186789842 cites W2804905867 @default.
- W3186789842 cites W2804935296 @default.
- W3186789842 cites W2901026139 @default.
- W3186789842 cites W2901394229 @default.
- W3186789842 cites W2906598409 @default.
- W3186789842 cites W2910603373 @default.
- W3186789842 cites W2910745941 @default.
- W3186789842 cites W2911801153 @default.
- W3186789842 cites W2963460174 @default.
- W3186789842 cites W2963474687 @default.
- W3186789842 cites W2963703197 @default.
- W3186789842 cites W2963813458 @default.
- W3186789842 cites W2964186069 @default.
- W3186789842 cites W2964350391 @default.
- W3186789842 cites W2995472160 @default.
- W3186789842 cites W2995653869 @default.
- W3186789842 cites W3003448987 @default.
- W3186789842 cites W3015045227 @default.
- W3186789842 cites W3015146382 @default.
- W3186789842 cites W3034978746 @default.
- W3186789842 cites W3035003500 @default.
- W3186789842 doi "https://doi.org/10.48550/arxiv.2107.08585" @default.
- W3186789842 hasPublicationYear "2021" @default.
- W3186789842 type Work @default.
- W3186789842 sameAs 3186789842 @default.
- W3186789842 citedByCount "0" @default.
- W3186789842 crossrefType "posted-content" @default.
- W3186789842 hasAuthorship W3186789842A5030379402 @default.
- W3186789842 hasAuthorship W3186789842A5063135412 @default.
- W3186789842 hasAuthorship W3186789842A5063366658 @default.
- W3186789842 hasBestOaLocation W31867898421 @default.
- W3186789842 hasConcept C111919701 @default.
- W3186789842 hasConcept C115961682 @default.
- W3186789842 hasConcept C119857082 @default.
- W3186789842 hasConcept C12267149 @default.
- W3186789842 hasConcept C127705205 @default.
- W3186789842 hasConcept C150899416 @default.
- W3186789842 hasConcept C153180895 @default.
- W3186789842 hasConcept C154945302 @default.
- W3186789842 hasConcept C2776135515 @default.
- W3186789842 hasConcept C33923547 @default.
- W3186789842 hasConcept C41008148 @default.
- W3186789842 hasConcept C48372109 @default.
- W3186789842 hasConcept C66905080 @default.
- W3186789842 hasConcept C8642999 @default.
- W3186789842 hasConcept C94375191 @default.
- W3186789842 hasConceptScore W3186789842C111919701 @default.
- W3186789842 hasConceptScore W3186789842C115961682 @default.
- W3186789842 hasConceptScore W3186789842C119857082 @default.
- W3186789842 hasConceptScore W3186789842C12267149 @default.
- W3186789842 hasConceptScore W3186789842C127705205 @default.
- W3186789842 hasConceptScore W3186789842C150899416 @default.
- W3186789842 hasConceptScore W3186789842C153180895 @default.
- W3186789842 hasConceptScore W3186789842C154945302 @default.
- W3186789842 hasConceptScore W3186789842C2776135515 @default.
- W3186789842 hasConceptScore W3186789842C33923547 @default.
- W3186789842 hasConceptScore W3186789842C41008148 @default.
- W3186789842 hasConceptScore W3186789842C48372109 @default.
- W3186789842 hasConceptScore W3186789842C66905080 @default.
- W3186789842 hasConceptScore W3186789842C8642999 @default.
- W3186789842 hasConceptScore W3186789842C94375191 @default.
- W3186789842 hasLocation W31867898421 @default.
- W3186789842 hasOpenAccess W3186789842 @default.
- W3186789842 hasPrimaryLocation W31867898421 @default.
- W3186789842 hasRelatedWork W3172364442 @default.
- W3186789842 hasRelatedWork W3186789842 @default.
- W3186789842 hasRelatedWork W3215867059 @default.
- W3186789842 hasRelatedWork W4210794429 @default.
- W3186789842 hasRelatedWork W4223456145 @default.
- W3186789842 hasRelatedWork W4280535922 @default.
- W3186789842 hasRelatedWork W4295309597 @default.
- W3186789842 hasRelatedWork W4304128395 @default.
- W3186789842 hasRelatedWork W4308262314 @default.
- W3186789842 hasRelatedWork W4309113015 @default.
- W3186789842 isParatext "false" @default.