Abstract: A semiconductor device includes a shielding wire formed across a semiconductor die and an auxiliary wire supporting the shielding wire, thereby reducing the size of a package while shielding the electromagnetic interference generated from the semiconductor die. In one embodiment, the semiconductor device includes a substrate having at least one circuit device mounted thereon, a semiconductor die spaced apart from the circuit device and mounted on the substrate, a shielding wire spaced apart from the semiconductor die and formed across the semiconductor die, and an auxiliary wire supporting the shielding wire under the shielding wire and formed to be perpendicular to the shielding wire. In another embodiment, a bump structure is used to support the shielding wire. In a further embodiment, an auxiliary wire includes a bump structure portion and wire portion and both the bump structure portion and the wire portion are used to support the shielding wire.

Abstract: The present disclosure relates to a method and an apparatus for federated learning of an artificial intelligence model. According to an exemplary embodiment of the present disclosure, a federated learning method of an artificial intelligence model includes: training a first local artificial intelligence model and a second local artificial intelligence model using data sets of a first client and a second client among the plurality of clients; calculating performance values for the first local artificial intelligence model and the second local artificial intelligence model by transmitting the first local artificial intelligence model to the second client and transmitting the second local artificial intelligence model to the first client; comparing the performance values to remove one of the first client and the second client; and training a global model using a client which is not removed.

Abstract: A local model training method of a federated learning framework implementing training data classification is provided. In the local model training method, a client may classify training data into two categories, generate a learning mini-batch by adjusting a ratio between samples classified into the two categories and included in the mini-batch to a preset ratio, and train a learning model using the mini-batch with the adjusted sample ratio.

Abstract: Provided is a method of removing, by a server, an abnormal client in federated learning. A method of removing, by a server, an abnormal client in federated learning may include receiving, from a user equipment (UE), first weight values trained in a first local model, generating a first client model based on the first weight values, validating the first client model by using a validation data set in order to determine whether the first client model is legitimate, and removing the first weight values based on the first client model not being legitimate.

Abstract: A shoe management device has a housing forming an exterior appearance, an inner space provided inside of a housing body of the housing, and a cover to open or close the housing body. A shoe may be provided inside of the inner space, and dust and foreign matter on the shoe may be removed by agitators and an air flow provided by an air generator. In addition, moisture and germs inside the shoe are removed by hot air, an ultraviolet ray, and plasma ions provided by a care device.

Abstract: A shoe management device may have a housing having a body to define an inner space and a cover to open or close the body to allow access to the inner space. Agitators may be provided at a bottom of the body, which may be configured to hold shoes, and the agitators may remove foreign matter from the shoes. An upper portion of the body may have an air generator configured to blow air toward the shoes to remove foreign matter. In addition, moisture and germs inside the shoe may be removed by hot air, ultraviolet light, and plasma ions provided by a care assembly provided inside of the body and configured to be at least partially inserted into the shoes.

Abstract: Disclosed are a bone conduction speaker module and bone conduction earphones having the bone conduction speaker modules. More particularly, the present invention relates to a bone conduction speaker module and bone conduction earphones having the bone conduction speaker modules, wherein the bone conduction speaker modules can accurately output sound, can stably output sound with fewer vibrations, enable a wearer to comfortably use the earphones, can prevent strain to the wearer by minimizing influence of magnetic force, can facilitate wearing of the earphones, and can prevent an accident from occurring by allowing the wearer to hear external sounds. To this end, a bone conduction speaker module used in earphones, headphones, and a hearing aid, includes: a housing including a first casing covering a portion coming into contact with a human body and a second casing covering a side of the first casing; a magnet part; and a voice coil part.