sps44 commited on
Commit
edaa503
1 Parent(s): 21555d0

Updated readme

Browse files
Files changed (1) hide show
  1. README.md +21 -17
README.md CHANGED
@@ -73,7 +73,7 @@ Once a year, the [DCASE community](https://dcase.community/) publishes a [challe
73
 
74
  ### Explore Dataset
75
 
76
- ![Analyze DCASE23 Task 2 with Spotlight](data/preview_dcase.png "Analyze DCASE23 Task 2 with Spotlight")
77
 
78
  The enrichments allow you to quickly gain insights into the dataset. The open source data curation tool Renumics Spotlight enables that with just a few lines of code:
79
 
@@ -84,9 +84,9 @@ import datasets
84
  dataset = datasets.load_dataset("renumics/dcase23-task2-enriched", "dev", split="all", streaming=False)
85
  ```
86
 
87
- Install Spotlight via [pip](https://packaging.python.org/en/latest/key_projects/#pip):
88
  ```jupyterpython
89
- !pip install renumics-spotlight
90
  ```
91
 
92
  Start exploring with a simple view that leverages embeddings to identify relevant data segments:
@@ -94,40 +94,37 @@ Start exploring with a simple view that leverages embeddings to identify relevan
94
  from renumics import spotlight
95
 
96
  df = dataset.to_pandas()
97
- spotlight.show(df, dtype={'path': spotlight.Audio, "embeddings_ast-finetuned-audioset-10-10-0.4593": spotlight.Embedding})
 
98
  ```
99
  You can use the UI to interactively configure the view on the data. Depending on the concrete taks (e.g. model comparison, debugging, outlier detection) you might want to leverage different enrichments and metadata.
100
 
101
- This example shows a view that is suitable to compare the results of two different ML models:
102
 
103
 
104
  ```jupyterpython
105
  from renumics import spotlight
106
 
107
- df = dataset.to_pandas()
108
- # spotlight.show(df, dtype={'path': spotlight.Audio, "embeddings_ast-finetuned-audioset-10-10-0.4593": spotlight.Embedding}, layout=ds_builder.config.get_layout(config="standard"))
109
- # todo
110
  ```
111
 
112
  ## Using custom model results and enrichments
113
 
114
  When developing your custom model you want to use different kinds of information from you model (e.g. embedding, anomaly scores etc.) to gain further insights into the dataset and the model behvior.
115
 
116
- Suppose you have your model's embeddings for each datapoint as a 2D-Numpy array and your anomaly score as a 1D-Numpy array. Then you can add this information to the dataset:
117
  ```jupyterpython
118
- from renumics import spotlight
119
- #todo
120
  ```
121
  Depending on your concrete task you might want to use different enrichments. For a good overview on great open source tooling for uncertainty quantification, explainability and outlier detection, you can take a look at our [curated list for open source data-centric AI tooling](https://github.com/Renumics/awesome-open-data-centric-ai) on Github.
122
 
123
  You can also save your view configuration in Spotlight in a JSON configuration file by clicking on the respective icon:
124
 
125
- [//]: # (todo: create new image -> Spotlight widgets use column "audio" as key in this picture.)
126
-
127
  ![Save a data curation layout in Spotlight](data/spotlight_save_layout.png "Save a data curation layout in Spotlight")
128
 
129
- We have also set up a Github repo with interactive notebooks that highlight different enrichment possibilities and curation templates. We encourage you to submit a pull request to share your enrichment code and interaction layouts.
130
-
131
 
132
  ## Dataset Structure
133
 
@@ -136,8 +133,6 @@ We have also set up a Github repo with interactive notebooks that highlight diff
136
  For each instance, there is a Audio for the audio, a string for the path, an integer for the section, a string for the d1p (parameter), a string for the d1v (value),
137
  a ClassLabel for the label and a ClassLabel for the class.
138
 
139
- [//]: # (todo: add new data fields)
140
-
141
  ```
142
  {'audio': {'array': array([ 0. , 0.00024414, -0.00024414, ..., -0.00024414,
143
  0. , 0. ], dtype=float32),
@@ -158,6 +153,10 @@ a ClassLabel for the label and a ClassLabel for the class.
158
  'ast-finetuned-audioset-10-10-0.4593-embeddings': [0.8152204155921936,
159
  1.5862374305725098, ...,
160
  1.7154160737991333]
 
 
 
 
161
  }
162
  ```
163
 
@@ -176,6 +175,11 @@ The length of each audio file is 10 seconds.
176
  - `class`: an integer as class label.
177
  - `label`: an integer whose value may be either _0_, indicating that the audio sample is _normal_, _1_, indicating that the audio sample contains an _anomaly_.
178
  - `ast-finetuned-audioset-10-10-0.4593-embeddings`: an `datasets.Sequence(Value("float32"))` representing audio embeddings that are generated with an [Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer#transformers.ASTFeatureExtractor).
 
 
 
 
 
179
 
180
  ### Data Splits
181
 
 
73
 
74
  ### Explore Dataset
75
 
76
+ ![Analyze DCASE23 Task 2 with Spotlight](data/preview_dcase_1.png "Analyze DCASE23 Task 2 with Spotlight")
77
 
78
  The enrichments allow you to quickly gain insights into the dataset. The open source data curation tool Renumics Spotlight enables that with just a few lines of code:
79
 
 
84
  dataset = datasets.load_dataset("renumics/dcase23-task2-enriched", "dev", split="all", streaming=False)
85
  ```
86
 
87
+ Install soundfile and Spotlight via [pip](https://packaging.python.org/en/latest/key_projects/#pip):
88
  ```jupyterpython
89
+ !pip install renumics-spotlight soundfile
90
  ```
91
 
92
  Start exploring with a simple view that leverages embeddings to identify relevant data segments:
 
94
  from renumics import spotlight
95
 
96
  df = dataset.to_pandas()
97
+ simple_layout = load_dataset_builder("renumics/dcase23-task2-enriched", "dev").get_layout(config="simple")
98
+ spotlight.show(df, dtype={'path': spotlight.Audio, "embeddings_ast-finetuned-audioset-10-10-0.4593": spotlight.Embedding}, layout=simple_layout)
99
  ```
100
  You can use the UI to interactively configure the view on the data. Depending on the concrete taks (e.g. model comparison, debugging, outlier detection) you might want to leverage different enrichments and metadata.
101
 
102
+ In this example we focus on the valve class. We specifically look at normal data points that have high anomaly scores in both models. This is one example on how to find difficult example or edge cases:
103
 
104
 
105
  ```jupyterpython
106
  from renumics import spotlight
107
 
108
+ extended_layout = load_dataset_builder("renumics/dcase23-task2-enriched", "dev").get_layout(config="extended")
109
+ spotlight.show(df, dtype={'path': spotlight.Audio, "embeddings_ast-finetuned-audioset-10-10-0.4593": spotlight.Embedding}, layout=extended_layout)
 
110
  ```
111
 
112
  ## Using custom model results and enrichments
113
 
114
  When developing your custom model you want to use different kinds of information from you model (e.g. embedding, anomaly scores etc.) to gain further insights into the dataset and the model behvior.
115
 
116
+ Suppose you have your model's embeddings for each datapoint as a 2D-Numpy array called `embeddings` and your anomaly score as a 1D-Numpy array called `anomaly_scores`. Then you can add this information to the dataset:
117
  ```jupyterpython
118
+ df['my_model_embedding'] = embeddings
119
+ df['anomaly_score'] = anomaly_score
120
  ```
121
  Depending on your concrete task you might want to use different enrichments. For a good overview on great open source tooling for uncertainty quantification, explainability and outlier detection, you can take a look at our [curated list for open source data-centric AI tooling](https://github.com/Renumics/awesome-open-data-centric-ai) on Github.
122
 
123
  You can also save your view configuration in Spotlight in a JSON configuration file by clicking on the respective icon:
124
 
 
 
125
  ![Save a data curation layout in Spotlight](data/spotlight_save_layout.png "Save a data curation layout in Spotlight")
126
 
127
+ For more information how to configure the Spotlight UI please refer to the [documentation](https://spotlight.renumics.com).
 
128
 
129
  ## Dataset Structure
130
 
 
133
  For each instance, there is a Audio for the audio, a string for the path, an integer for the section, a string for the d1p (parameter), a string for the d1v (value),
134
  a ClassLabel for the label and a ClassLabel for the class.
135
 
 
 
136
  ```
137
  {'audio': {'array': array([ 0. , 0.00024414, -0.00024414, ..., -0.00024414,
138
  0. , 0. ], dtype=float32),
 
153
  'ast-finetuned-audioset-10-10-0.4593-embeddings': [0.8152204155921936,
154
  1.5862374305725098, ...,
155
  1.7154160737991333]
156
+ 'anomaly_score_dcase2023_task2_baseline_ae': 8.284389
157
+ 'prediction_dcase2023_task2_baseline_ae': 0
158
+ 'prediction_correct_dcase2023_task2_baseline_ae': 1
159
+ 'anomaly_score_embedding_lof': 0.191043
160
  }
161
  ```
162
 
 
175
  - `class`: an integer as class label.
176
  - `label`: an integer whose value may be either _0_, indicating that the audio sample is _normal_, _1_, indicating that the audio sample contains an _anomaly_.
177
  - `ast-finetuned-audioset-10-10-0.4593-embeddings`: an `datasets.Sequence(Value("float32"))` representing audio embeddings that are generated with an [Audio Spectrogram Transformer](https://huggingface.co/docs/transformers/model_doc/audio-spectrogram-transformer#transformers.ASTFeatureExtractor).
178
+ - `anomaly_score_dcase2023_task2_baseline_ae`: a float representation of the anomaly score according to the baseline implementation
179
+ - `prediction_dcase2023_task2_baseline_ae`: an integer whose value may be either _0_, indicating that the audio sample is considered _normal_ by the baseline algorithm, _1_, indicating that the audio sample contains an _anomaly_.
180
+ - `prediction_correct_dcase2023_task2_baseline_ae`: an integer whose value may be either _0_, indicating that the baseline prediction is wrong or _1_, indicating that prediction is correct.
181
+ - `anomaly_score_embedding_lof`: a float representation of the anomaly score computed with the PyOD implementation of the Local Outlier Factor algorithm on the pre-computed embedding.
182
+
183
 
184
  ### Data Splits
185