feature map
In plain wordsThe grid of numbers produced when one learned pattern detector is scanned across an image, showing where, and how strongly, that pattern appears.
Convolutional networks
The output of one filterfilter, n.: In an image-recognition network, a small grid of learned numbers that acts as a pattern detector, for vertical edges, a patch of fur or a wheel, scanned across the whole picture. of a convolutionalconvolution, n.: An operation that slides a small grid of weights across an image or signal, multiplying and adding at each spot, to measure how strongly a pattern appears at every location. layer over all spatial positions: a 2-D array, for images, recording how strongly the filter's pattern is present at each location. A layer's output is a stack of feature maps, one per output channel, usually stored as a tensor of shape channels × height × width. Early feature maps keep fine spatial detail; deeper ones are coarser but encode more abstract patterns, which is why detection and segmentation architectures such as feature pyramid networks and U-NetU-Net, prop. n.: An image network shaped like the letter U: one half shrinks the picture to understand it, the other enlarges it back, with shortcuts between the halves so fine detail is not lost. combine maps from several depths. Feature maps of pretrained networks are reused for transfer learningtransfer learning, n.: Reusing what a model learned on one task to get a head start on another — like a pianist picking up the organ faster than a beginner. It lets models learn new jobs from little data., style transferstyle transfer, n.: Repainting a photo in the style of an artwork — your street as Van Gogh might have painted it — by keeping the photo's content and borrowing the painting's colors, strokes and textures., perceptual losses and class-activation heat maps.
Upsample the last feature map and overlay it on the image to see which regions drove the prediction.
Origin
Used in early convolutional-network papers: Yann LeCun and colleagues' 1989 zip-code network arranged its hidden units as independent 'feature maps' sharing weights, and the term was standard by the LeNet-5 paper (1998). Kunihiko Fukushima's neocognitron (1980) had analogous 'cell planes'.
Usage note
Distinct from the feature map φ of kernel methodskernel trick, n.: A mathematical shortcut that lets simple straight-line methods find curved patterns, by computing similarities between examples as if the data had been mapped into a far richer space — without ever building that space., which maps inputs into a feature space, and from the transformer-interpretability notion of a featurefeature (interpretability), n.: A concept or property that a neural network represents internally — such as 'is French', 'mentions a bridge' or 'this code has a bug' — usually corresponding to a direction in its internal activity. as a direction in activation space.
Related
- filterfilter, n.: In an image-recognition network, a small grid of learned numbers that acts as a pattern detector, for vertical edges, a patch of fur or a wheel, scanned across the whole picture.
- convolutionconvolution, n.: An operation that slides a small grid of weights across an image or signal, multiplying and adding at each spot, to measure how strongly a pattern appears at every location.
- convolutional neural networkconvolutional neural network, n.: A kind of neural network built for images and other grid-like data. It slides small pattern detectors across the input, finding edges first, then shapes, then whole objects.
- receptive fieldreceptive field, n.: The patch of the original input, such as a region of an image, that a particular neuron inside a network can see and respond to.
- activationactivation, n.: The numbers a neural network computes inside itself as it processes an input — a reading of which of its artificial neurons are firing, and how strongly, at that moment.
- poolingpooling, n.: A step in an image network that shrinks its internal picture by summarizing each small patch with one number, usually the largest or the average, keeping what matters and discarding exact positions.
Mentioned in
backbonebackbone, n.: The main, usually pretrained part of a vision model that turns an image into useful internal features; smaller task-specific parts are attached to it, like limbs to a spine., convolutionconvolution, n.: An operation that slides a small grid of weights across an image or signal, multiplying and adding at each spot, to measure how strongly a pattern appears at every location., convolutional neural networkconvolutional neural network, n.: A kind of neural network built for images and other grid-like data. It slides small pattern detectors across the input, finding edges first, then shapes, then whole objects., featurefeature, n.: One piece of information about each example that a model uses, like a house's size or a pixel's color, or a pattern a model has learned to detect inside itself., filterfilter, n.: In an image-recognition network, a small grid of learned numbers that acts as a pattern detector, for vertical edges, a patch of fur or a wheel, scanned across the whole picture., Grad-CAMGrad-CAM, abbr.: A technique that shows, as a coarse heat map, which regions of an image a vision network relied on when choosing a particular label., kernelkernel, n.: A word with several AI meanings: a similarity measure used by some learning methods; a small grid of numbers that scans an image inside a neural network; or a small program that runs on a GPU., poolingpooling, n.: A step in an image network that shrinks its internal picture by summarizing each small patch with one number, usually the largest or the average, keeping what matters and discarding exact positions., receptive fieldreceptive field, n.: The patch of the original input, such as a region of an image, that a particular neuron inside a network can see and respond to., stridestride, n.: How far a scanning window jumps each time it moves across an image. Jumping two pixels instead of one halves the size of the result., U-NetU-Net, prop. n.: An image network shaped like the letter U: one half shrinks the picture to understand it, the other enlarges it back, with shortcuts between the halves so fine detail is not lost., anchor boxanchor box, n.: A ready-made box shape, placed at many spots across an image, that an object detector stretches and shifts to fit real objects instead of drawing boxes from scratch., DenseNetDenseNet, prop. n.: An image-recognition network in which every layer receives the outputs of all the layers before it, so features are reused rather than relearned., equivarianceequivariance, n.: A property of a model whose output changes in step with its input: move the cat in a photo, and the network's internal 'cat' signal moves the same way., fully convolutional networkfully convolutional network, n.: An image network made only of scanning-window layers, with no fixed-size final layers, so it can take pictures of any size and label every pixel., perceptual lossperceptual loss, n.: A way of judging how different two images look to a person by comparing what a trained image-recognition network 'sees' in each, instead of comparing pixels one by one., squeeze-and-excitation networksqueeze-and-excitation network, n.: An image-network design whose add-on blocks let the network decide, for each picture, which of its internal feature channels to turn up and which to turn down.