YOU SAID:

Deep learning has, in relatively few years, improved significantly the performance of many machine learning applications. Even though its popularity has surged, it's not always easy to apply it to a real-world problem. Developing a good deep learning model is a process that most likely will include several iterations of data collection, hyperparameter tuning and training. One big obstacle is its hunger for data and compute power. For supervised learning one often requires a massive amount of annotated examples, and training usually extends over hours or days. This makes the process of applying deep learning very time and resource consuming. We investigate if recent advances in few-shot learning can be used to speed up this process, and look specifically at object detection as an example. Such methods could potentially both decrease the necessary number of examples and the training time. MAML \autocite{MAML} is a promising few-shot learning method based on meta-learning that optimizes the initial parameters of a model to be best possibly suited for fine-tuning. It's model-agnostic in nature and can in principle be applied to most deep learning models. But through extensive exploration we show that it's far from trivial to apply MAML to object detection on natural images. Instead, we are able to use a simpler method inspired by MAML, Reptile \autocite{Reptile2}. We show that a model pretrained using Reptile can be fine-tuned considerably faster than a model pretrained normally on object detection, but surprisingly it does not enable using fewer examples. In addition, we show that Reptile is actually able to speed up the development of deep learning models in practice. This is done by building a proof of concept tool and use this to perform some example use cases.

INTO JAPANESE

ディープラーニングは、比較的数年で多くの機械学習アプリケーションのパフォーマンスを大幅に向上させました。人気が高まっているにもかかわらず、現実の問題にそれを適用することは必ずしも容易ではありません。良い深い学習モデルを開発することは、データ収集のいくつかの反復、ハイパーパラメー

BACK INTO ENGLISH

Deplaning the significantly improved performance of many machine learning applications in relatively few years. Applying it to real-world problems are becoming increasingly popular, even though is not necessarily easy. To develop a good deep learning model

INTO JAPANESE

比較的数年の間に、多くの機械学習アプリケーションの大幅に改善された性能を解き放つ。実際の問題にそれを適用することは、必ずしも容易ではないが、ますます普及している。良い深い学習モデルを開発する

BACK INTO ENGLISH

In a relatively few years, unleash greatly improved performance of many machine learning applications. Applying it to real problems is not always easy, but it is becoming more and more popular. Develop a good deep learning model

INTO JAPANESE

比較的数年の間に、多くの機械学習アプリケーションのパフォーマンスが大幅に向上しました。実際の問題に適用することは必ずしも容易ではありませんが、ますます普及しています。良い深い学習モデルを開発する

BACK INTO ENGLISH

Within a few years, the performance of many machine learning applications has improved significantly. It is not always easy to apply to actual problems, but it is getting more and more popular. Develop a good deep learning model

INTO JAPANESE

数年内で多くのマシン学習アプリケーションのパフォーマンスは著しく良くなりました。簡単じゃない常に実際の問題に適用するより多くの人気になっています。良い深い学習モデルを開発します。

BACK INTO ENGLISH

Several years in many machine learning applications performance is significantly improved. Has become the most popular application of real problem always not easy. Develop a good deep learning model.

INTO JAPANESE

数年の機械学習アプリケーションでは、パフォーマンスが大幅に向上しました。実際の問題の最も普及したアプリケーションになっているのは常に容易ではありません。良い深い学習モデルを開発する。

BACK INTO ENGLISH

In the machine learning application of several years, the performance has improved greatly. It is not always easy to become the most popular application of real problems. Develop a good deep learning model.

INTO JAPANESE

数年の機械学習アプリケーションでは、パフォーマンスが大幅に向上しました。実際の問題の最も普及したアプリケーションになることは必ずしも容易ではありません。良い深い学習モデルを開発する。

BACK INTO ENGLISH

In the machine learning application of several years, the performance has improved greatly. It is not always easy to become the most popular application of actual problems. Develop a good deep learning model.

INTO JAPANESE

数年の機械学習アプリケーションでは、パフォーマンスが大幅に向上しました。実際の問題の最も一般的なアプリケーションになることは必ずしも容易ではありません。良い深い学習モデルを開発する。

BACK INTO ENGLISH

In the machine learning application of several years, the performance has improved greatly. It is not always easy to become the most common application of real problems. Develop a good deep learning model.

INTO JAPANESE

数年の機械学習アプリケーションでは、パフォーマンスが大幅に向上しました。実際の問題の最も一般的なアプリケーションになることは必ずしも容易ではありません。良い深い学習モデルを開発する。

BACK INTO ENGLISH

In the machine learning application of several years, the performance has improved greatly. It is not always easy to become the most common application of real problems. Develop a good deep learning model.