![]() (Left) T-SNE visualization of the embed- ding of tasksĮxtracted from the iNaturalist, CUB-200, iMaterialist datasets.Ĭolors indicate ground-truth grouping of tasks based on taxonomic We use this to select an expert from a given collection,įigure 1: Task embedding across a large library of tasks (best seen Model embedding, called MODEL2VEC, in such a way that models whoseĮmbeddings are close to a task exhibit good perfmormance on the To address this, we learn a joint task and The task, and ignores interactions with the model which may however Insufficient data to train or fine- tune a generic model, and This can be particu- larly valuable when there is Problem of selecting the best pre-trained feature extractor to Our task embedding can be used to reason about the space of tasksĪnd solve meta-tasks. “difficulty” of the task, character- istics of the input domain,Īnd which features of the probe network are useful to solve it Network are fixed, the FIM provides a fixed-dimensional Since the architecture and weights of the probe Of the network filter parameters to capture the structure of the Work”, and compute the diagonal Fisher Information Ma- trix (FIM) Reference convolutional neural network which we call “probe net. Of labeled samples, we feed the data through a pre-trained Specifically, given a task defined by a dataset D = Ni=1 Computation of the embedding leverages a duality be- tween network parameters (weights) and outputs (activa- tions) in a deep neural network (DNN): Just as the activa- tions of a DNN trained on a complex visual recognition task are a rich representation of the input images, we show that the gradients of the weights relative to a task-specific loss are a rich representation of the task itself. More- over, we introduce an asymmetric distance on tasks which correlates with the transferability between tasks. When other natural distances are available, such as the taxonomical dis- tance in biological classification, we find that the embed- ding distance correlates positively with it (Fig. The norm of the embedding correlates with the complexity of the task, while the distance between embeddings captures semantic similarities between tasks (Fig. We introduce the TASK 2 VEC embedding, a technique to represent tasks as elements of a vector space based on the Fisher Information Matrix. Yet, no general framework exists to describe and learn relations between tasks. Introduction The success of Deep Learning hinges in part on the fact that models learned for one task can be used on other related tasks. Se- lecting a feature extractor with task embedding obtains a performance close to the best available feature extractor, while costing substantially less than exhaustively training and evaluating on all available feature extractors. We present a simple meta-learning frame- work for learning a metric on embeddings that is capable of predicting which feature extractors will perform well. We also demonstrate the practical value of this framework for the meta-task of selecting a pre-trained feature extractor for a new task. , tasks based on classifying different types of plants are similar). ![]() We demon- strate that this embedding is capable of predicting task sim- ilarities that match our intuition about semantic and tax- onomic relations between different visual tasks ( e.g. This provides a fixed-dimensional embedding of the task that is independent of details such as the number of classes and does not require any understanding of the class label semantics. Given a dataset with ground-truth labels and a loss function defined over those labels, we process images through a “probe network” and compute an embedding based on estimates of the Fisher information matrix asso- ciated with the probe network parameters. T ASK 2V EC : Task Embedding for Meta-Learning Alessandro Achille UCLA and AWS Michael Lam AWS Rahul Tewari AWS Avinash Ravichandran AWS Subhransu Maji UMass and AWS Charless Fowlkes UCI and AWS Stefano Soatto UCLA and AWS Pietro Perona Caltech and AWS Abstract We introduce a method to provide vectorial represen- tations of visual classification tasks which can be used to reason about the nature of those tasks and their re- lations. ![]()
0 Comments
Leave a Reply. |